chore: bump dashboard to v0.2.1

chore: bump version to v0.2.0
chore: fix clippy (#1387 )
2025-12-22 22:20:02 +00:00 · 2023-04-18 14:32:07 +08:00 · 2023-04-17 15:16:49 +08:00 · 2023-04-15 07:00:54 +08:00 · 2023-04-15 00:03:15 +08:00 · 2023-04-14 22:26:37 +08:00
629 changed files with 40960 additions and 14098 deletions
--- a/.cargo/config.toml
+++ b/.cargo/config.toml
@@ -3,3 +3,13 @@ linker = "aarch64-linux-gnu-gcc"

 [alias]
 sqlness = "run --bin sqlness-runner --"
+
+
+[build]
+rustflags = [
+    # lints
+    # TODO: use lint configuration in cargo https://github.com/rust-lang/cargo/issues/5034
+    "-Wclippy::print_stdout",
+    "-Wclippy::print_stderr",
+    "-Wclippy::implicit_clone",
+]
--- a/.env.example
+++ b/.env.example
@@ -2,7 +2,7 @@
 GT_S3_BUCKET=S3 bucket
 GT_S3_ACCESS_KEY_ID=S3 access key id
 GT_S3_ACCESS_KEY=S3 secret access key
-
+GT_S3_ENDPOINT_URL=S3 endpoint url
 # Settings for oss test
 GT_OSS_BUCKET=OSS bucket
 GT_OSS_ACCESS_KEY_ID=OSS access key id
--- a/.github/pr-title-breaking-change-label-config.json
+++ b/.github/pr-title-breaking-change-label-config.json
@@ -0,0 +1,13 @@
+{
+    "LABEL": {
+        "name": "breaking change",
+        "color": "D93F0B"
+    },
+    "CHECKS": {
+        "regexp": "^(?:(?!!:).)*$",
+        "ignoreLabels": [
+            "ignore-title"
+        ],
+        "alwaysPassCI": true
+    }
+}
--- a/.github/pr-title-checker-config.json
+++ b/.github/pr-title-checker-config.json
@@ -1,10 +1,12 @@
 {
-  "LABEL": {
-    "name": "Invalid PR Title",
-    "color": "B60205"
-  },
-  "CHECKS": {
-    "regexp": "^(feat|fix|test|refactor|chore|style|docs|perf|build|ci|revert)(\\(.*\\))?:.*",
-    "ignoreLabels" : ["ignore-title"]
-  }
+    "LABEL": {
+        "name": "Invalid PR Title",
+        "color": "B60205"
+    },
+    "CHECKS": {
+        "regexp": "^(feat|fix|test|refactor|chore|style|docs|perf|build|ci|revert)(\\(.*\\))?\\!?:.*",
+        "ignoreLabels": [
+            "ignore-title"
+        ]
+    }
 }
--- a/.github/workflows/apidoc.yml
+++ b/.github/workflows/apidoc.yml
@@ -0,0 +1,42 @@
+on:
+  push:
+    branches:
+      - develop
+    paths-ignore:
+      - 'docs/**'
+      - 'config/**'
+      - '**.md'
+      - '.dockerignore'
+      - 'docker/**'
+      - '.gitignore'
+
+name: Build API docs
+
+env:
+  RUST_TOOLCHAIN: nightly-2023-02-26
+
+jobs:
+  apidoc:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v3
+    - uses: arduino/setup-protoc@v1
+      with:
+        repo-token: ${{ secrets.GITHUB_TOKEN }}
+    - uses: dtolnay/rust-toolchain@master
+      with:
+        toolchain: ${{ env.RUST_TOOLCHAIN }}
+    - run: cargo doc --workspace --no-deps --document-private-items
+    - run: |
+        cat <<EOF > target/doc/index.html
+        <!DOCTYPE html>
+        <html>
+        <head>
+        <meta http-equiv="refresh" content="0; url='greptime/'" />
+        </head>
+        <body></body></html>
+        EOF
+    - name: Publish dist directory
+      uses: JamesIves/github-pages-deploy-action@v4
+      with:
+        folder: target/doc
--- a/.github/workflows/develop.yml
+++ b/.github/workflows/develop.yml
@@ -24,15 +24,15 @@ on:
 name: CI

 env:
-  RUST_TOOLCHAIN: nightly-2023-02-14
+  RUST_TOOLCHAIN: nightly-2023-02-26

 jobs:
  typos:
    name: Spell Check with Typos
    runs-on: ubuntu-latest
    steps:
-      - uses: actions/checkout@v2
-      - uses: crate-ci/typos@v1.0.4
+      - uses: actions/checkout@v3
+      - uses: crate-ci/typos@v1.13.10

  check:
    name: Check
@@ -116,6 +116,7 @@ jobs:
    if: github.event.pull_request.draft == false
    runs-on: ubuntu-latest-8-cores
    timeout-minutes: 60
+    needs: [clippy]
    steps:
      - uses: actions/checkout@v3
      - uses: arduino/setup-protoc@v1
@@ -131,7 +132,7 @@ jobs:
          ETCD_VER=v3.5.7
          DOWNLOAD_URL=https://github.com/etcd-io/etcd/releases/download
          curl -L ${DOWNLOAD_URL}/${ETCD_VER}/etcd-${ETCD_VER}-linux-amd64.tar.gz -o /tmp/etcd-${ETCD_VER}-linux-amd64.tar.gz
-          mkdir -p /tmp/etcd-download 
+          mkdir -p /tmp/etcd-download
          tar xzvf /tmp/etcd-${ETCD_VER}-linux-amd64.tar.gz -C /tmp/etcd-download --strip-components=1
          rm -f /tmp/etcd-${ETCD_VER}-linux-amd64.tar.gz

@@ -182,12 +183,13 @@ jobs:
      - name: Rust Cache
        uses: Swatinem/rust-cache@v2
      - name: Run cargo clippy
-        run: cargo clippy --workspace --all-targets -- -D warnings -D clippy::print_stdout -D clippy::print_stderr
+        run: cargo clippy --workspace --all-targets -- -D warnings

  coverage:
    if: github.event.pull_request.draft == false
    runs-on: ubuntu-latest-8-cores
    timeout-minutes: 60
+    needs: [clippy]
    steps:
      - uses: actions/checkout@v3
      - uses: arduino/setup-protoc@v1
@@ -205,10 +207,16 @@ jobs:
        uses: Swatinem/rust-cache@v2
      - name: Install latest nextest release
        uses: taiki-e/install-action@nextest
+      - name: Install Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.10'
+      - name: Install PyArrow Package
+        run: pip install pyarrow
      - name: Install cargo-llvm-cov
        uses: taiki-e/install-action@cargo-llvm-cov
      - name: Collect coverage data
-        run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info
+        run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info -F pyo3_backend
        env:
          CARGO_BUILD_RUSTFLAGS: "-C link-arg=-fuse-ld=lld"
          RUST_BACKTRACE: 1
@@ -223,5 +231,5 @@ jobs:
          token: ${{ secrets.CODECOV_TOKEN }}
          files: ./lcov.info
          flags: rust
-          fail_ci_if_error: true
+          fail_ci_if_error: false
          verbose: true
--- a/.github/workflows/license.yaml
+++ b/.github/workflows/license.yaml
@@ -13,4 +13,4 @@ jobs:
    steps:
    - uses: actions/checkout@v2
    - name: Check License Header
-      uses: apache/skywalking-eyes/header@main
+      uses: apache/skywalking-eyes/header@df70871af1a8109c9a5b1dc824faaf65246c5236
--- a/.github/workflows/pr-title-checker.yml
+++ b/.github/workflows/pr-title-checker.yml
@@ -18,3 +18,12 @@ jobs:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          pass_on_octokit_error: false
          configuration_path: ".github/pr-title-checker-config.json"
+  breaking:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - uses: thehanimo/pr-title-checker@v1.3.4
+        with:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          pass_on_octokit_error: false
+          configuration_path: ".github/pr-title-breaking-change-label-config.json"
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -5,20 +5,22 @@ on:
  schedule:
    # At 00:00 on Monday.
    - cron: '0 0 * * 1'
+  # Mannually trigger only builds binaries.
  workflow_dispatch:

 name: Release

 env:
-  RUST_TOOLCHAIN: nightly-2023-02-14
+  RUST_TOOLCHAIN: nightly-2023-02-26

-  # FIXME(zyy17): Would be better to use `gh release list -L 1 | cut -f 3` to get the latest release version tag, but for a long time, we will stay at 'v0.1.0-alpha-*'.
-  SCHEDULED_BUILD_VERSION_PREFIX: v0.1.0-alpha
+  SCHEDULED_BUILD_VERSION_PREFIX: v0.2.0

-  # In the future, we can change SCHEDULED_PERIOD to nightly.
-  SCHEDULED_PERIOD: weekly
+  SCHEDULED_PERIOD: nightly

-  CARGO_PROFILE: weekly
+  CARGO_PROFILE: nightly
+
+  ## FIXME(zyy17): Enable it after the tests are stabled.
+  DISABLE_RUN_TESTS: true

 jobs:
  build:
@@ -30,16 +32,45 @@ jobs:
          - arch: x86_64-unknown-linux-gnu
            os: ubuntu-2004-16-cores
            file: greptime-linux-amd64
+            continue-on-error: false
+            opts: "-F servers/dashboard"
          - arch: aarch64-unknown-linux-gnu
            os: ubuntu-2004-16-cores
            file: greptime-linux-arm64
+            continue-on-error: false
+            opts: "-F servers/dashboard"
          - arch: aarch64-apple-darwin
            os: macos-latest
            file: greptime-darwin-arm64
+            continue-on-error: false
+            opts: "-F servers/dashboard"
          - arch: x86_64-apple-darwin
            os: macos-latest
            file: greptime-darwin-amd64
+            continue-on-error: false
+            opts: "-F servers/dashboard"
+          - arch: x86_64-unknown-linux-gnu
+            os: ubuntu-2004-16-cores
+            file: greptime-linux-amd64-pyo3
+            continue-on-error: false
+            opts: "-F pyo3_backend,servers/dashboard"
+          - arch: aarch64-unknown-linux-gnu
+            os: ubuntu-2004-16-cores
+            file: greptime-linux-arm64-pyo3
+            continue-on-error: false
+            opts: "-F pyo3_backend,servers/dashboard"
+          - arch: aarch64-apple-darwin
+            os: macos-latest
+            file: greptime-darwin-arm64-pyo3
+            continue-on-error: false
+            opts: "-F pyo3_backend,servers/dashboard"
+          - arch: x86_64-apple-darwin
+            os: macos-latest
+            file: greptime-darwin-amd64-pyo3
+            continue-on-error: false
+            opts: "-F pyo3_backend,servers/dashboard"
    runs-on: ${{ matrix.os }}
+    continue-on-error: ${{ matrix.continue-on-error }}
    if: github.repository == 'GreptimeTeam/greptimedb'
    steps:
      - name: Checkout sources
@@ -76,10 +107,10 @@ jobs:
          ETCD_VER=v3.5.7
          DOWNLOAD_URL=https://github.com/etcd-io/etcd/releases/download
          curl -L ${DOWNLOAD_URL}/${ETCD_VER}/etcd-${ETCD_VER}-linux-amd64.tar.gz -o /tmp/etcd-${ETCD_VER}-linux-amd64.tar.gz
-          mkdir -p /tmp/etcd-download 
-          tar xzvf /tmp/etcd-${ETCD_VER}-linux-amd64.tar.gz -C /tmp/etcd-download --strip-components=1 
+          mkdir -p /tmp/etcd-download
+          tar xzvf /tmp/etcd-${ETCD_VER}-linux-amd64.tar.gz -C /tmp/etcd-download --strip-components=1
          rm -f /tmp/etcd-${ETCD_VER}-linux-amd64.tar.gz
-          
+
          sudo cp -a /tmp/etcd-download/etcd* /usr/local/bin/
          nohup etcd >/tmp/etcd.log 2>&1 &

@@ -93,7 +124,14 @@ jobs:
        if: contains(matrix.arch, 'linux') && endsWith(matrix.arch, '-gnu')
        run: |
          sudo apt-get -y update
-          sudo apt-get -y install libssl-dev pkg-config g++-aarch64-linux-gnu gcc-aarch64-linux-gnu
+          sudo apt-get -y install libssl-dev pkg-config g++-aarch64-linux-gnu gcc-aarch64-linux-gnu binutils-aarch64-linux-gnu wget
+
+      # FIXME(zyy17): Should we specify the version of python when building binary for darwin?
+      - name: Compile Python 3.10.10 from source for linux
+        if: contains(matrix.arch, 'linux') && contains(matrix.opts, 'pyo3_backend')
+        run: |
+          sudo chmod +x ./docker/aarch64/compile-python.sh
+          sudo ./docker/aarch64/compile-python.sh ${{ matrix.arch }}

      - name: Install rust toolchain
        uses: dtolnay/rust-toolchain@master
@@ -105,10 +143,55 @@ jobs:
        run: protoc --version ; cargo version ; rustc --version ; gcc --version ; g++ --version

      - name: Run tests
+        if: env.DISABLE_RUN_TESTS == 'false'
        run: make unit-test integration-test sqlness-test

+      - name: Run cargo build with pyo3 for aarch64-linux
+        if: contains(matrix.arch, 'aarch64-unknown-linux-gnu') && contains(matrix.opts, 'pyo3_backend')
+        run: |
+          # TODO(zyy17): We should make PYO3_CROSS_LIB_DIR configurable.
+          export PYTHON_INSTALL_PATH_AMD64=${PWD}/python-3.10.10/amd64
+          export LD_LIBRARY_PATH=$PYTHON_INSTALL_PATH_AMD64/lib:$LD_LIBRARY_PATH
+          export LIBRARY_PATH=$PYTHON_INSTALL_PATH_AMD64/lib:$LIBRARY_PATH
+          export PATH=$PYTHON_INSTALL_PATH_AMD64/bin:$PATH
+
+          export PYO3_CROSS_LIB_DIR=${PWD}/python-3.10.10/aarch64
+          echo "PYO3_CROSS_LIB_DIR: $PYO3_CROSS_LIB_DIR"
+          alias python=$PYTHON_INSTALL_PATH_AMD64/bin/python3
+          alias pip=$PYTHON_INSTALL_PATH_AMD64/bin/python3-pip
+
+          cargo build --profile ${{ env.CARGO_PROFILE }} --locked --target ${{ matrix.arch }} ${{ matrix.opts }}
+
+      - name: Run cargo build with pyo3 for amd64-linux
+        if: contains(matrix.arch, 'x86_64-unknown-linux-gnu') && contains(matrix.opts, 'pyo3_backend')
+        run: |
+          export PYTHON_INSTALL_PATH_AMD64=${PWD}/python-3.10.10/amd64
+          export LD_LIBRARY_PATH=$PYTHON_INSTALL_PATH_AMD64/lib:$LD_LIBRARY_PATH
+          export LIBRARY_PATH=$PYTHON_INSTALL_PATH_AMD64/lib:$LIBRARY_PATH
+          export PATH=$PYTHON_INSTALL_PATH_AMD64/bin:$PATH
+
+          echo "implementation=CPython" >> pyo3.config
+          echo "version=3.10" >> pyo3.config
+          echo "implementation=CPython" >> pyo3.config
+          echo "shared=true" >> pyo3.config
+          echo "abi3=true" >> pyo3.config
+          echo "lib_name=python3.10" >> pyo3.config
+          echo "lib_dir=$PYTHON_INSTALL_PATH_AMD64/lib" >> pyo3.config
+          echo "executable=$PYTHON_INSTALL_PATH_AMD64/bin/python3" >> pyo3.config
+          echo "pointer_width=64" >> pyo3.config
+          echo "build_flags=" >> pyo3.config
+          echo "suppress_build_script_link_lines=false" >> pyo3.config
+
+          cat pyo3.config
+          export PYO3_CONFIG_FILE=${PWD}/pyo3.config
+          alias python=$PYTHON_INSTALL_PATH_AMD64/bin/python3
+          alias pip=$PYTHON_INSTALL_PATH_AMD64/bin/python3-pip
+
+          cargo build --profile ${{ env.CARGO_PROFILE }} --locked --target ${{ matrix.arch }} ${{ matrix.opts }}
+
      - name: Run cargo build
-        run: cargo build ${{ matrix.opts }} --profile ${{ env.CARGO_PROFILE }} --locked --target ${{ matrix.arch }}
+        if: contains(matrix.arch, 'darwin') || contains(matrix.opts, 'pyo3_backend') == false
+        run: cargo build --profile ${{ env.CARGO_PROFILE }} --locked --target ${{ matrix.arch }} ${{ matrix.opts }}

      - name: Calculate checksum and rename binary
        shell: bash
@@ -129,89 +212,16 @@ jobs:
        with:
          name: ${{ matrix.file }}.sha256sum
          path: target/${{ matrix.arch }}/${{ env.CARGO_PROFILE }}/${{ matrix.file }}.sha256sum
-  release:
-    name: Release artifacts
-    needs: [build]
-    runs-on: ubuntu-latest
-    if: github.repository == 'GreptimeTeam/greptimedb'
-    steps:
-      - name: Checkout sources
-        uses: actions/checkout@v3
-
-      - name: Download artifacts
-        uses: actions/download-artifact@v3
-
-      - name: Configure scheduled build version # the version would be ${SCHEDULED_BUILD_VERSION_PREFIX}-YYYYMMDD-${SCHEDULED_PERIOD}, like v0.1.0-alpha-20221119-weekly.
-        shell: bash
-        if: github.event_name == 'schedule'
-        run: |
-          buildTime=`date "+%Y%m%d"`
-          SCHEDULED_BUILD_VERSION=${{ env.SCHEDULED_BUILD_VERSION_PREFIX }}-$buildTime-${{ env.SCHEDULED_PERIOD }}
-          echo "SCHEDULED_BUILD_VERSION=${SCHEDULED_BUILD_VERSION}" >> $GITHUB_ENV
-
-      - name: Create scheduled build git tag
-        if: github.event_name == 'schedule'
-        run: |
-          git tag ${{ env.SCHEDULED_BUILD_VERSION }}
-
-      - name: Publish scheduled release # configure the different release title and tags.
-        uses: softprops/action-gh-release@v1
-        if: github.event_name == 'schedule'
-        with:
-          name: "Release ${{ env.SCHEDULED_BUILD_VERSION }}"
-          tag_name: ${{ env.SCHEDULED_BUILD_VERSION }}
-          generate_release_notes: true
-          files: |
-            **/greptime-*
-
-      - name: Publish release
-        uses: softprops/action-gh-release@v1
-        if: github.event_name != 'schedule'
-        with:
-          name: "Release ${{ github.ref_name }}"
-          files: |
-            **/greptime-*

  docker:
    name: Build docker image
    needs: [build]
    runs-on: ubuntu-latest
-    if: github.repository == 'GreptimeTeam/greptimedb'
+    if: github.repository == 'GreptimeTeam/greptimedb' && github.event_name != 'workflow_dispatch'
    steps:
      - name: Checkout sources
        uses: actions/checkout@v3

-      - name: Download amd64 binary
-        uses: actions/download-artifact@v3
-        with:
-          name: greptime-linux-amd64
-          path: amd64
-
-      - name: Unzip the amd64 artifacts
-        run: |
-          cd amd64
-          tar xvf greptime-linux-amd64.tgz
-          rm greptime-linux-amd64.tgz
-
-      - name: Download arm64 binary
-        uses: actions/download-artifact@v3
-        with:
-          name: greptime-linux-arm64
-          path: arm64
-
-      - name: Unzip the arm64 artifacts
-        run: |
-          cd arm64
-          tar xvf greptime-linux-arm64.tgz
-          rm greptime-linux-arm64.tgz
-
-      - name: Login to UCloud Container Registry
-        uses: docker/login-action@v2
-        with:
-          registry: uhub.service.ucloud.cn
-          username: ${{ secrets.UCLOUD_USERNAME }}
-          password: ${{ secrets.UCLOUD_PASSWORD }}
-
      - name: Login to Dockerhub
        uses: docker/login-action@v2
        with:
@@ -239,15 +249,141 @@ jobs:
      - name: Set up buildx
        uses: docker/setup-buildx-action@v2

-      - name: Build and push
-        uses: docker/build-push-action@v3
+      - name: Download amd64 binary
+        uses: actions/download-artifact@v3
        with:
-          context: .
+          name: greptime-linux-amd64-pyo3
+          path: amd64
+
+      - name: Unzip the amd64 artifacts
+        run: |
+          tar xvf amd64/greptime-linux-amd64-pyo3.tgz -C amd64/ && rm amd64/greptime-linux-amd64-pyo3.tgz
+          cp -r amd64 docker/ci
+
+      - name: Download arm64 binary
+        id: download-arm64
+        uses: actions/download-artifact@v3
+        with:
+          name: greptime-linux-arm64-pyo3
+          path: arm64
+
+      - name: Unzip the arm64 artifacts
+        id: unzip-arm64
+        if: success() || steps.download-arm64.conclusion == 'success'
+        run: |
+          tar xvf arm64/greptime-linux-arm64-pyo3.tgz -C arm64/ && rm arm64/greptime-linux-arm64-pyo3.tgz
+          cp -r arm64 docker/ci
+
+      - name: Build and push all
+        uses: docker/build-push-action@v3
+        if: success() || steps.unzip-arm64.conclusion == 'success' # Build and push all platform if unzip-arm64 succeeds
+        with:
+          context: ./docker/ci/
          file: ./docker/ci/Dockerfile
          push: true
          platforms: linux/amd64,linux/arm64
          tags: |
            greptime/greptimedb:latest
            greptime/greptimedb:${{ env.IMAGE_TAG }}
-            uhub.service.ucloud.cn/greptime/greptimedb:latest
-            uhub.service.ucloud.cn/greptime/greptimedb:${{ env.IMAGE_TAG }}
+
+      - name: Build and push amd64 only
+        uses: docker/build-push-action@v3
+        if: success() || steps.download-arm64.conclusion == 'failure' # Only build and push amd64 platform if download-arm64 fails
+        with:
+          context: ./docker/ci/
+          file: ./docker/ci/Dockerfile
+          push: true
+          platforms: linux/amd64
+          tags: |
+            greptime/greptimedb:latest
+            greptime/greptimedb:${{ env.IMAGE_TAG }}
+
+  release:
+    name: Release artifacts
+    # Release artifacts only when all the artifacts are built successfully.
+    needs: [build,docker]
+    runs-on: ubuntu-latest
+    if: github.repository == 'GreptimeTeam/greptimedb' && github.event_name != 'workflow_dispatch'
+    steps:
+      - name: Checkout sources
+        uses: actions/checkout@v3
+
+      - name: Download artifacts
+        uses: actions/download-artifact@v3
+
+      - name: Configure scheduled build version # the version would be ${SCHEDULED_BUILD_VERSION_PREFIX}-${SCHEDULED_PERIOD}-YYYYMMDD, like v0.2.0-nigthly-20230313.
+        shell: bash
+        if: github.event_name == 'schedule'
+        run: |
+          buildTime=`date "+%Y%m%d"`
+          SCHEDULED_BUILD_VERSION=${{ env.SCHEDULED_BUILD_VERSION_PREFIX }}-${{ env.SCHEDULED_PERIOD }}-$buildTime
+          echo "SCHEDULED_BUILD_VERSION=${SCHEDULED_BUILD_VERSION}" >> $GITHUB_ENV
+
+      - name: Create scheduled build git tag
+        if: github.event_name == 'schedule'
+        run: |
+          git tag ${{ env.SCHEDULED_BUILD_VERSION }}
+
+      - name: Publish scheduled release # configure the different release title and tags.
+        uses: softprops/action-gh-release@v1
+        if: github.event_name == 'schedule'
+        with:
+          name: "Release ${{ env.SCHEDULED_BUILD_VERSION }}"
+          tag_name: ${{ env.SCHEDULED_BUILD_VERSION }}
+          generate_release_notes: true
+          files: |
+            **/greptime-*
+
+      - name: Publish release
+        uses: softprops/action-gh-release@v1
+        if: github.event_name != 'schedule'
+        with:
+          name: "Release ${{ github.ref_name }}"
+          files: |
+            **/greptime-*
+
+  docker-push-uhub:
+    name: Push docker image to UCloud Container Registry
+    needs: [docker]
+    runs-on: ubuntu-latest
+    if: github.repository == 'GreptimeTeam/greptimedb' && github.event_name != 'workflow_dispatch'
+    # Push to uhub may fail(500 error), but we don't want to block the release process. The failed job will be retried manually.
+    continue-on-error: true
+    steps:
+      - name: Checkout sources
+        uses: actions/checkout@v3
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v2
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+
+      - name: Login to UCloud Container Registry
+        uses: docker/login-action@v2
+        with:
+          registry: uhub.service.ucloud.cn
+          username: ${{ secrets.UCLOUD_USERNAME }}
+          password: ${{ secrets.UCLOUD_PASSWORD }}
+
+      - name: Configure scheduled build image tag # the tag would be ${SCHEDULED_BUILD_VERSION_PREFIX}-YYYYMMDD-${SCHEDULED_PERIOD}
+        shell: bash
+        if: github.event_name == 'schedule'
+        run: |
+          buildTime=`date "+%Y%m%d"`
+          SCHEDULED_BUILD_VERSION=${{ env.SCHEDULED_BUILD_VERSION_PREFIX }}-$buildTime-${{ env.SCHEDULED_PERIOD }}
+          echo "IMAGE_TAG=${SCHEDULED_BUILD_VERSION:1}" >> $GITHUB_ENV
+
+      - name: Configure tag # If the release tag is v0.1.0, then the image version tag will be 0.1.0.
+        shell: bash
+        if: github.event_name != 'schedule'
+        run: |
+          VERSION=${{ github.ref_name }}
+          echo "IMAGE_TAG=${VERSION:1}" >> $GITHUB_ENV
+
+      - name: Push image to uhub # Use 'docker buildx imagetools create' to create a new image base on source image.
+        run: |
+          docker buildx imagetools create \
+            --tag uhub.service.ucloud.cn/greptime/greptimedb:latest \
+            --tag uhub.service.ucloud.cn/greptime/greptimedb:${{ env.IMAGE_TAG }} \
+            greptime/greptimedb:${{ env.IMAGE_TAG }}
--- a/.gitignore
+++ b/.gitignore
@@ -35,3 +35,7 @@ benchmarks/data

 # dotenv
 .env
+
+# dashboard files
+!/src/servers/dashboard/VERSION
+/src/servers/dashboard/*
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -1,4 +1,4 @@
-# Welcome!
+# Welcome 👋

 Thanks a lot for considering contributing to GreptimeDB. We believe people like you would make GreptimeDB a great product. We intend to build a community where individuals can have open talks, show respect for one another, and speak with true ❤️. Meanwhile, we are to keep transparency and make your effort count here.

@@ -50,34 +50,33 @@ GreptimeDB uses the [Apache 2.0 license](https://github.com/GreptimeTeam/greptim

 - To ensure that community is free and confident in its ability to use your contributions, please sign the Contributor License Agreement (CLA) which will be incorporated in the pull request process.
 - Make sure all your codes are formatted and follow the [coding style](https://pingcap.github.io/style-guide/rust/).
- Make sure all unit tests are passed.
- Make sure all clippy warnings are fixed (you can check it locally by running `cargo clippy --workspace --all-targets -- -D warnings -D clippy::print_stdout -D clippy::print_stderr`).
+- Make sure all unit tests are passed (using `cargo test --workspace` or [nextest](https://nexte.st/index.html) `cargo nextest run`).
+- Make sure all clippy warnings are fixed (you can check it locally by running `cargo clippy --workspace --all-targets -- -D warnings`).

 #### `pre-commit` Hooks
+
 You could setup the [`pre-commit`](https://pre-commit.com/#plugins) hooks to run these checks on every commit automatically.

 1. Install `pre-commit`
-```
-$ pip install pre-commit
-```
-or
-```
-$ brew install pre-commit
-```
+
+        pip install pre-commit
+
+    or
+
+        brew install pre-commit

 2. Install the `pre-commit` hooks
-```
-$ pre-commit install
-pre-commit installed at .git/hooks/pre-commit

-$ pre-commit install --hook-type commit-msg
-pre-commit installed at .git/hooks/commit-msg
+        $ pre-commit install
+        pre-commit installed at .git/hooks/pre-commit

-$ pre-commit install --hook-type pre-push
-pre-commit installed at .git/hooks/pre-pus
-```
+        $ pre-commit install --hook-type commit-msg
+        pre-commit installed at .git/hooks/commit-msg

-now `pre-commit` will run automatically on `git commit`.
+        $ pre-commit install --hook-type pre-push
+        pre-commit installed at .git/hooks/pre-push
+
+Now, `pre-commit` will run automatically on `git commit`.

 ### Title

@@ -102,10 +101,12 @@ of what you were trying to do and what went wrong. You can also reach for help i
 ## Community

 The core team will be thrilled if you participate in any way you like. When you are stuck, try ask for help by filing an issue, with a detailed description of what you were trying to do and what went wrong. If you have any questions or if you would like to get involved in our community, please check out:
+
 - [GreptimeDB Community Slack](https://greptime.com/slack)
 - [GreptimeDB Github Discussions](https://github.com/GreptimeTeam/greptimedb/discussions)

 Also, see some extra GreptimeDB content:
+
 - [GreptimeDB Docs](https://greptime.com/docs)
 - [Learn GreptimeDB](https://greptime.com/products/db)
 - [Greptime Inc. Website](https://greptime.com)
--- a/Cargo.lock
+++ b/Cargo.lock
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -7,20 +7,24 @@ members = [
    "src/cmd",
    "src/common/base",
    "src/common/catalog",
+    "src/common/datasource",
    "src/common/error",
    "src/common/function",
    "src/common/function-macro",
    "src/common/grpc",
    "src/common/grpc-expr",
+    "src/common/mem-prof",
    "src/common/procedure",
    "src/common/query",
    "src/common/recordbatch",
    "src/common/runtime",
    "src/common/substrait",
    "src/common/telemetry",
+    "src/common/test-util",
    "src/common/time",
    "src/datanode",
    "src/datatypes",
+    "src/file-table-engine",
    "src/frontend",
    "src/log-store",
    "src/meta-client",
@@ -37,48 +41,50 @@ members = [
    "src/storage",
    "src/store-api",
    "src/table",
+    "src/table-procedure",
    "tests-integration",
    "tests/runner",
 ]

 [workspace.package]
-version = "0.1.0"
+version = "0.2.0"
 edition = "2021"
 license = "Apache-2.0"

 [workspace.dependencies]
-arrow = "29.0"
-arrow-array = "29.0"
-arrow-flight = "29.0"
-arrow-schema = { version = "29.0", features = ["serde"] }
+arrow = { version = "37.0" }
+arrow-array = "37.0"
+arrow-flight = "37.0"
+arrow-schema = { version = "37.0", features = ["serde"] }
 async-stream = "0.3"
 async-trait = "0.1"
 chrono = { version = "0.4", features = ["serde"] }
-# TODO(LFC): Use released Datafusion when it officially dpendent on Arrow 29.0
-datafusion = { git = "https://github.com/apache/arrow-datafusion.git", rev = "4917235a398ae20145c87d20984e6367dc1a0c1e" }
-datafusion-common = { git = "https://github.com/apache/arrow-datafusion.git", rev = "4917235a398ae20145c87d20984e6367dc1a0c1e" }
-datafusion-expr = { git = "https://github.com/apache/arrow-datafusion.git", rev = "4917235a398ae20145c87d20984e6367dc1a0c1e" }
-datafusion-optimizer = { git = "https://github.com/apache/arrow-datafusion.git", rev = "4917235a398ae20145c87d20984e6367dc1a0c1e" }
-datafusion-physical-expr = { git = "https://github.com/apache/arrow-datafusion.git", rev = "4917235a398ae20145c87d20984e6367dc1a0c1e" }
-datafusion-sql = { git = "https://github.com/apache/arrow-datafusion.git", rev = "4917235a398ae20145c87d20984e6367dc1a0c1e" }
+datafusion = { git = "https://github.com/apache/arrow-datafusion.git", rev = "74a778ca6016a853a3c3add3fa8c6f12f4fe4561" }
+datafusion-common = { git = "https://github.com/apache/arrow-datafusion.git", rev = "74a778ca6016a853a3c3add3fa8c6f12f4fe4561" }
+datafusion-expr = { git = "https://github.com/apache/arrow-datafusion.git", rev = "74a778ca6016a853a3c3add3fa8c6f12f4fe4561" }
+datafusion-optimizer = { git = "https://github.com/apache/arrow-datafusion.git", rev = "74a778ca6016a853a3c3add3fa8c6f12f4fe4561" }
+datafusion-physical-expr = { git = "https://github.com/apache/arrow-datafusion.git", rev = "74a778ca6016a853a3c3add3fa8c6f12f4fe4561" }
+datafusion-sql = { git = "https://github.com/apache/arrow-datafusion.git", rev = "74a778ca6016a853a3c3add3fa8c6f12f4fe4561" }
 futures = "0.3"
 futures-util = "0.3"
-parquet = "29.0"
+parquet = "37.0"
 paste = "1.0"
 prost = "0.11"
+rand = "0.8"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 snafu = { version = "0.7", features = ["backtraces"] }
-sqlparser = "0.28"
+sqlparser = "0.33"
+tempfile = "3"
 tokio = { version = "1.24.2", features = ["full"] }
-tokio-util = "0.7"
-tonic = { version = "0.8", features = ["tls"] }
+tokio-util = { version = "0.7", features = ["io-util"] }
+tonic = { version = "0.9", features = ["tls"] }
 uuid = { version = "1", features = ["serde", "v4", "fast-rng"] }

 [profile.release]
 debug = true

-[profile.weekly]
+[profile.nightly]
 inherits = "release"
 strip = true
 lto = "thin"
--- a/10
+++ b/10
@@ -19,6 +19,14 @@ clean: ## Clean the project.
 fmt: ## Format all the Rust code.
 	cargo fmt --all

+.PHONY: fmt-toml
+fmt-toml: ## Format all TOML files.
+	taplo format --option "indent_string=    "
+
+.PHONY: check-toml
+check-toml: ## Check all TOML files.
+	taplo format --check --option "indent_string=    "
+
 .PHONY: docker-image
 docker-image: ## Build docker image.
 	docker build --network host -f docker/Dockerfile -t ${IMAGE_REGISTRY}:${IMAGE_TAG} .
@@ -43,7 +51,7 @@ check: ## Cargo check all the targets.

 .PHONY: clippy
 clippy: ## Check clippy rules.
-	cargo clippy --workspace --all-targets -- -D warnings -D clippy::print_stdout -D clippy::print_stderr
+	cargo clippy --workspace --all-targets -- -D warnings

 .PHONY: fmt-check
 fmt-check: ## Check code format.
--- a/README.md
+++ b/README.md
@@ -1,14 +1,14 @@
 <p align="center">
  <picture>
-    <source media="(prefers-color-scheme: light)" srcset="/docs/logo-text-padding.png">
-    <source media="(prefers-color-scheme: dark)" srcset="/docs/logo-text-padding-dark.png">
-    <img alt="GreptimeDB Logo" src="/docs/logo-text-padding.png" width="400px">
+    <source media="(prefers-color-scheme: light)" srcset="https://cdn.jsdelivr.net/gh/GreptimeTeam/greptimedb@develop/docs/logo-text-padding.png">
+    <source media="(prefers-color-scheme: dark)" srcset="https://cdn.jsdelivr.net/gh/GreptimeTeam/greptimedb@develop/docs/logo-text-padding-dark.png">
+    <img alt="GreptimeDB Logo" src="https://cdn.jsdelivr.net/gh/GreptimeTeam/greptimedb@develop/docs/logo-text-padding.png" width="400px">
  </picture>
 </p>


 <h3 align="center">
-    The next-generation hybrid timeseries/analytics processing database in the cloud
+    The next-generation hybrid time-series/analytics processing database in the cloud
 </h3>

 <p align="center">
@@ -23,6 +23,8 @@
    <a href="https://twitter.com/greptime"><img src="https://img.shields.io/badge/twitter-follow_us-1d9bf0.svg"></a>
    &nbsp;
    <a href="https://www.linkedin.com/company/greptime/"><img src="https://img.shields.io/badge/linkedin-connect_with_us-0a66c2.svg"></a>
+    &nbsp;
+    <a href="https://greptime.com/slack"><img src="https://img.shields.io/badge/slack-GreptimeDB-0abd59?logo=slack" alt="slack" /></a>
 </p>

 ## What is GreptimeDB
@@ -36,11 +38,11 @@ Our core developers have been building time-series data platform
 for years. Based on their best-practices, GreptimeDB is born to give you:

 - A standalone binary that scales to highly-available distributed cluster, providing a transparent experience for cluster users
- Optimized columnar layout for handling time-series data; compacted, compressed, stored on various storage backends
- Flexible index options, tackling high cardinality issues down
+- Optimized columnar layout for handling time-series data; compacted, compressed, and stored on various storage backends
+- Flexible indexes, tackling high cardinality issues down
 - Distributed, parallel query execution, leveraging elastic computing resource
 - Native SQL, and Python scripting for advanced analytical scenarios
- Widely adopted database protocols and APIs
+- Widely adopted database protocols and APIs, native PromQL supports
 - Extensible table engine architecture for extensive workloads

 ## Quick Start
@@ -61,6 +63,12 @@ To compile GreptimeDB from source, you'll need:
  find an installation instructions [here](https://grpc.io/docs/protoc-installation/).
  **Note that `protoc` version needs to be >= 3.15** because we have used the `optional`
  keyword. You can check it with `protoc --version`.
+- python3-dev or python3-devel(Optional feature, only needed if you want to run scripts
+  in CPython, and also need to enable `pyo3_backend` feature when compiling(by `cargo run -F pyo3_backend` or add `pyo3_backend` to src/script/Cargo.toml 's `features.default` like `default = ["python", "pyo3_backend]`)): this install a Python shared library required for running Python
+  scripting engine(In CPython Mode). This is available as `python3-dev` on
+  ubuntu, you can install it with `sudo apt install python3-dev`, or
+  `python3-devel` on RPM based distributions (e.g. Fedora, Red Hat, SuSE). Mac's
+  `Python3` package should have this shared library by default. More detail for compiling with PyO3 can be found in [PyO3](https://pyo3.rs/v0.18.1/building_and_distribution#configuring-the-python-version)'s documentation.

 #### Build with Docker

@@ -141,9 +149,9 @@ You can always cleanup test database by removing `/tmp/greptimedb`.
 ### Installation

 - [Pre-built Binaries](https://github.com/GreptimeTeam/greptimedb/releases):
-  downloadable pre-built binaries for Linux and MacOS
- [Docker Images](https://hub.docker.com/r/greptime/greptimedb): pre-built
-  Docker images
+  For Linux and macOS, you can easily download pre-built binaries that are ready to use. In most cases, downloading the version without PyO3 is sufficient. However, if you plan to run scripts in CPython (and use Python packages like NumPy and Pandas), you will need to download the version with PyO3 and install a Python with the same version as the Python in the PyO3 version. We recommend using virtualenv for the installation process to manage multiple Python versions.
+- [Docker Images](https://hub.docker.com/r/greptime/greptimedb)(**recommended**): pre-built
+  Docker images, this is the easiest way to try GreptimeDB. By default it runs CPython script with `pyo3_backend` enabled.
 - [`gtctl`](https://github.com/GreptimeTeam/gtctl): the command-line tool for
  Kubernetes deployment

@@ -152,6 +160,7 @@ You can always cleanup test database by removing `/tmp/greptimedb`.
 - GreptimeDB [User Guide](https://docs.greptime.com/user-guide/concepts.html)
 - GreptimeDB [Developer
  Guide](https://docs.greptime.com/developer-guide/overview.html)
+- GreptimeDB [internal code document](https://greptimedb.rs)

 ### Dashboard
 - [The dashboard UI for GreptimeDB](https://github.com/GreptimeTeam/dashboard)
--- a/SECURITY.md
+++ b/SECURITY.md
@@ -0,0 +1,19 @@
+# Security Policy
+
+## Supported Versions
+
+| Version | Supported          |
+| ------- | ------------------ |
+| >= v0.1.0 | :white_check_mark: |
+| < v0.1.0   | :x:                |
+
+## Reporting a Vulnerability
+
+We place great importance on the security of GreptimeDB code, software, 
+and cloud platform. If you come across a security vulnerability in GreptimeDB, 
+we kindly request that you inform us immediately. We will thoroughly investigate
+all valid reports and make every effort to resolve the issue promptly.
+
+To report any issues or vulnerabilities, please email us at info@greptime.com, rather than 
+posting publicly on GitHub. Be sure to provide us with the version identifier as well as details
+on how the vulnerability can be exploited.
--- a/benchmarks/src/bin/nyc-taxi.rs
+++ b/benchmarks/src/bin/nyc-taxi.rs
@@ -21,13 +21,13 @@ use std::collections::HashMap;
 use std::path::{Path, PathBuf};
 use std::time::Instant;

-use arrow::array::{ArrayRef, PrimitiveArray, StringArray, TimestampNanosecondArray};
+use arrow::array::{ArrayRef, PrimitiveArray, StringArray, TimestampMicrosecondArray};
 use arrow::datatypes::{DataType, Float64Type, Int64Type};
 use arrow::record_batch::RecordBatch;
 use clap::Parser;
 use client::api::v1::column::Values;
-use client::api::v1::{Column, ColumnDataType, ColumnDef, CreateTableExpr, InsertRequest, TableId};
-use client::{Client, Database};
+use client::api::v1::{Column, ColumnDataType, ColumnDef, CreateTableExpr, InsertRequest};
+use client::{Client, Database, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
 use indicatif::{MultiProgress, ProgressBar, ProgressStyle};
 use parquet::arrow::arrow_reader::ParquetRecordBatchReaderBuilder;
 use tokio::task::JoinSet;
@@ -61,7 +61,7 @@ struct Args {
    #[arg(long = "skip-read")]
    skip_read: bool,

-    #[arg(short, long, default_value_t = String::from("127.0.0.1:3001"))]
+    #[arg(short, long, default_value_t = String::from("127.0.0.1:4001"))]
    endpoint: String,
 }

@@ -97,6 +97,9 @@ async fn write_data(

    for record_batch in record_batch_reader {
        let record_batch = record_batch.unwrap();
+        if !is_record_batch_full(&record_batch) {
+            continue;
+        }
        let (columns, row_count) = convert_record_batch(record_batch);
        let request = InsertRequest {
            table_name: TABLE_NAME.to_string(),
@@ -122,11 +125,17 @@ fn convert_record_batch(record_batch: RecordBatch) -> (Vec<Column>, u32) {
    let mut columns = vec![];

    for (array, field) in record_batch.columns().iter().zip(fields.iter()) {
-        let values = build_values(array);
+        let (values, datatype) = build_values(array);
+
        let column = Column {
-            column_name: field.name().to_owned(),
+            column_name: field.name().clone(),
            values: Some(values),
-            null_mask: vec![],
+            null_mask: array
+                .to_data()
+                .nulls()
+                .map(|bitmap| bitmap.buffer().as_slice().to_vec())
+                .unwrap_or_default(),
+            datatype: datatype.into(),
            // datatype and semantic_type are set to default
            ..Default::default()
        };
@@ -136,7 +145,7 @@ fn convert_record_batch(record_batch: RecordBatch) -> (Vec<Column>, u32) {
    (columns, row_count as _)
 }

-fn build_values(column: &ArrayRef) -> Values {
+fn build_values(column: &ArrayRef) -> (Values, ColumnDataType) {
    match column.data_type() {
        DataType::Int64 => {
            let array = column
@@ -144,10 +153,13 @@ fn build_values(column: &ArrayRef) -> Values {
                .downcast_ref::<PrimitiveArray<Int64Type>>()
                .unwrap();
            let values = array.values();
-            Values {
-                i64_values: values.to_vec(),
-                ..Default::default()
-            }
+            (
+                Values {
+                    i64_values: values.to_vec(),
+                    ..Default::default()
+                },
+                ColumnDataType::Int64,
+            )
        }
        DataType::Float64 => {
            let array = column
@@ -155,29 +167,38 @@ fn build_values(column: &ArrayRef) -> Values {
                .downcast_ref::<PrimitiveArray<Float64Type>>()
                .unwrap();
            let values = array.values();
-            Values {
-                f64_values: values.to_vec(),
-                ..Default::default()
-            }
+            (
+                Values {
+                    f64_values: values.to_vec(),
+                    ..Default::default()
+                },
+                ColumnDataType::Float64,
+            )
        }
        DataType::Timestamp(_, _) => {
            let array = column
                .as_any()
-                .downcast_ref::<TimestampNanosecondArray>()
+                .downcast_ref::<TimestampMicrosecondArray>()
                .unwrap();
            let values = array.values();
-            Values {
-                i64_values: values.to_vec(),
-                ..Default::default()
-            }
+            (
+                Values {
+                    ts_microsecond_values: values.to_vec(),
+                    ..Default::default()
+                },
+                ColumnDataType::TimestampMicrosecond,
+            )
        }
        DataType::Utf8 => {
            let array = column.as_any().downcast_ref::<StringArray>().unwrap();
            let values = array.iter().filter_map(|s| s.map(String::from)).collect();
-            Values {
-                string_values: values,
-                ..Default::default()
-            }
+            (
+                Values {
+                    string_values: values,
+                    ..Default::default()
+                },
+                ColumnDataType::String,
+            )
        }
        DataType::Null
        | DataType::Boolean
@@ -204,14 +225,19 @@ fn build_values(column: &ArrayRef) -> Values {
        | DataType::FixedSizeList(_, _)
        | DataType::LargeList(_)
        | DataType::Struct(_)
-        | DataType::Union(_, _, _)
+        | DataType::Union(_, _)
        | DataType::Dictionary(_, _)
        | DataType::Decimal128(_, _)
        | DataType::Decimal256(_, _)
+        | DataType::RunEndEncoded(_, _)
        | DataType::Map(_, _) => todo!(),
    }
 }

+fn is_record_batch_full(batch: &RecordBatch) -> bool {
+    batch.columns().iter().all(|col| col.null_count() == 0)
+}
+
 fn create_table_expr() -> CreateTableExpr {
    CreateTableExpr {
        catalog_name: CATALOG_NAME.to_string(),
@@ -227,13 +253,13 @@ fn create_table_expr() -> CreateTableExpr {
            },
            ColumnDef {
                name: "tpep_pickup_datetime".to_string(),
-                datatype: ColumnDataType::Int64 as i32,
+                datatype: ColumnDataType::TimestampMicrosecond as i32,
                is_nullable: true,
                default_constraint: vec![],
            },
            ColumnDef {
                name: "tpep_dropoff_datetime".to_string(),
-                datatype: ColumnDataType::Int64 as i32,
+                datatype: ColumnDataType::TimestampMicrosecond as i32,
                is_nullable: true,
                default_constraint: vec![],
            },
@@ -339,7 +365,8 @@ fn create_table_expr() -> CreateTableExpr {
        create_if_not_exists: false,
        table_options: Default::default(),
        region_ids: vec![0],
-        table_id: Some(TableId { id: 0 }),
+        table_id: None,
+        engine: "mito".to_string(),
    }
 }

@@ -422,7 +449,7 @@ fn main() {
        .unwrap()
        .block_on(async {
            let client = Client::with_urls(vec![&args.endpoint]);
-            let db = Database::with_client(client);
+            let db = Database::new(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, client);

            if !args.skip_write {
                do_write(&args, &db).await;
--- a/codecov.yml
+++ b/codecov.yml
@@ -8,3 +8,5 @@ coverage:
 ignore:
  - "**/error*.rs" # ignore all error.rs files
  - "tests/runner/*.rs" # ignore integration test runner
+comment:                  # this is a top-level key
+  layout: "diff"
--- a/config/datanode.example.toml
+++ b/config/datanode.example.toml
@@ -1,31 +1,60 @@
-node_id = 42
-mode = 'distributed'
-rpc_addr = '127.0.0.1:3001'
-rpc_hostname = '127.0.0.1'
-rpc_runtime_size = 8
-mysql_addr = '127.0.0.1:4406'
-mysql_runtime_size = 4
+# Node running mode, see `standalone.example.toml`.
+mode = "distributed"
+# Whether to use in-memory catalog, see `standalone.example.toml`.
 enable_memory_catalog = false
+# The datanode identifier, should be unique.
+node_id = 42
+# gRPC server address, "127.0.0.1:3001" by default.
+rpc_addr = "127.0.0.1:3001"
+# Hostname of this node.
+rpc_hostname = "127.0.0.1"
+# The number of gRPC server worker threads, 8 by default.
+rpc_runtime_size = 8

+# Metasrv client options.
+[meta_client_options]
+# Metasrv address list.
+metasrv_addrs = ["127.0.0.1:3002"]
+# Operation timeout in milliseconds, 3000 by default.
+timeout_millis = 3000
+# Connect server timeout in milliseconds, 5000 by default.
+connect_timeout_millis = 5000
+# `TCP_NODELAY` option for accepted connections, true by default.
+tcp_nodelay = true
+
+# WAL options, see `standalone.example.toml`.
 [wal]
 dir = "/tmp/greptimedb/wal"
-file_size = '1GB'
-purge_interval = '10m'
-purge_threshold = '50GB'
+file_size = "1GB"
+purge_threshold = "50GB"
+purge_interval = "10m"
 read_batch_size = 128
 sync_write = false

+# Storage options, see `standalone.example.toml`.
 [storage]
-type = 'File'
-data_dir = '/tmp/greptimedb/data/'
+type = "File"
+data_dir = "/tmp/greptimedb/data/"

-[meta_client_opts]
-metasrv_addrs = ['127.0.0.1:3002']
-timeout_millis = 3000
-connect_timeout_millis = 5000
-tcp_nodelay = false
-
-[compaction]
+# Compaction options, see `standalone.example.toml`.
+[storage.compaction]
 max_inflight_tasks = 4
-max_files_in_level0 = 16
+max_files_in_level0 = 8
 max_purge_tasks = 32
+
+# Storage manifest options
+[storage.manifest]
+# Region checkpoint actions margin.
+# Create a checkpoint every <checkpoint_margin> actions.
+checkpoint_margin = 10
+# Region manifest logs and checkpoints gc execution duration
+gc_duration = '30s'
+# Whether to try creating a manifest checkpoint on region opening
+checkpoint_on_startup = false
+
+# Procedure storage options, see `standalone.example.toml`.
+# [procedure.store]
+# type = "File"
+# data_dir = "/tmp/greptimedb/procedure/"
+# max_retry_times = 3
+# retry_delay = "500ms"
--- a/config/frontend.example.toml
+++ b/config/frontend.example.toml
@@ -1,12 +1,58 @@
-mode = 'distributed'
-datanode_rpc_addr = '127.0.0.1:3001'
+# Node running mode, see `standalone.example.toml`.
+mode = "distributed"

+# HTTP server options, see `standalone.example.toml`.
 [http_options]
-addr = '127.0.0.1:4000'
+addr = "127.0.0.1:4000"
 timeout = "30s"

-[meta_client_opts]
-metasrv_addrs = ['127.0.0.1:3002']
+# gRPC server options, see `standalone.example.toml`.
+[grpc_options]
+addr = "127.0.0.1:4001"
+runtime_size = 8
+
+# MySQL server options, see `standalone.example.toml`.
+[mysql_options]
+addr = "127.0.0.1:4002"
+runtime_size = 2
+
+# MySQL server TLS options, see `standalone.example.toml`.
+[mysql_options.tls]
+mode = "disable"
+cert_path = ""
+key_path = ""
+
+# PostgresSQL server options, see `standalone.example.toml`.
+[postgres_options]
+addr = "127.0.0.1:4003"
+runtime_size = 2
+
+# PostgresSQL server TLS options, see `standalone.example.toml`.
+[postgres_options.tls]
+mode = "disable"
+cert_path = ""
+key_path = ""
+
+# OpenTSDB protocol options, see `standalone.example.toml`.
+[opentsdb_options]
+addr = "127.0.0.1:4242"
+runtime_size = 2
+
+# InfluxDB protocol options, see `standalone.example.toml`.
+[influxdb_options]
+enable = true
+
+# Prometheus protocol options, see `standalone.example.toml`.
+[prometheus_options]
+enable = true
+
+# Prometheus protocol options, see `standalone.example.toml`.
+[prom_options]
+addr = "127.0.0.1:4004"
+
+# Metasrv client options, see `datanode.example.toml`.
+[meta_client_options]
+metasrv_addrs = ["127.0.0.1:3002"]
 timeout_millis = 3000
 connect_timeout_millis = 5000
-tcp_nodelay = false
+tcp_nodelay = true
--- a/config/metasrv.example.toml
+++ b/config/metasrv.example.toml
@@ -1,6 +1,15 @@
-bind_addr = '127.0.0.1:3002'
-server_addr = '127.0.0.1:3002'
-store_addr = '127.0.0.1:2379'
+# The bind address of metasrv, "127.0.0.1:3002" by default.
+bind_addr = "127.0.0.1:3002"
+# The communication server address for frontend and datanode to connect to metasrv,  "127.0.0.1:3002" by default for localhost.
+server_addr = "127.0.0.1:3002"
+# Etcd server address, "127.0.0.1:2379" by default.
+store_addr = "127.0.0.1:2379"
+# Datanode lease in seconds, 15 seconds by default.
 datanode_lease_secs = 15
-# selector: 'LeaseBased', 'LoadBased'
-selector = 'LeaseBased'
+# Datanode selector type.
+# - "LeaseBased" (default value).
+# - "LoadBased"
+# For details, please see "https://docs.greptime.com/developer-guide/meta/selector".
+selector = "LeaseBased"
+# Store data in memory, false by default.
+use_memory_store = false
--- a/config/standalone.example.toml
+++ b/config/standalone.example.toml
@@ -1,44 +1,130 @@
-node_id = 0
-mode = 'standalone'
+# Node running mode, "standalone" or "distributed".
+mode = "standalone"
+# Whether to use in-memory catalog, `false` by default.
 enable_memory_catalog = false

+# HTTP server options.
 [http_options]
-addr = '127.0.0.1:4000'
+# Server address, "127.0.0.1:4000" by default.
+addr = "127.0.0.1:4000"
+# HTTP request timeout, 30s by default.
 timeout = "30s"

-[wal]
-dir = "/tmp/greptimedb/wal"
-file_size = '1GB'
-purge_interval = '10m'
-purge_threshold = '50GB'
-read_batch_size = 128
-sync_write = false
-
-
-[storage]
-type = 'File'
-data_dir = '/tmp/greptimedb/data/'
-
+# gRPC server options.
 [grpc_options]
-addr = '127.0.0.1:4001'
+# Server address, "127.0.0.1:4001" by default.
+addr = "127.0.0.1:4001"
+# The number of server worker threads, 8 by default.
 runtime_size = 8

+# MySQL server options.
 [mysql_options]
-addr = '127.0.0.1:4002'
+# Server address, "127.0.0.1:4002" by default.
+addr = "127.0.0.1:4002"
+# The number of server worker threads, 2 by default.
 runtime_size = 2

-[influxdb_options]
-enable = true
-
-[opentsdb_options]
-addr = '127.0.0.1:4242'
-enable = true
-runtime_size = 2
-
-[prometheus_options]
-enable = true
+# MySQL server TLS options.
+[mysql_options.tls]
+# TLS mode, refer to https://www.postgresql.org/docs/current/libpq-ssl.html
+# - "disable" (default value)
+# - "prefer"
+# - "require"
+# - "verify-ca"
+# - "verify-full"
+mode = "disable"
+# Certificate file path.
+cert_path = ""
+# Private key file path.
+key_path = ""

+# PostgresSQL server options.
 [postgres_options]
-addr = '127.0.0.1:4003'
+# Server address, "127.0.0.1:4003" by default.
+addr = "127.0.0.1:4003"
+# The number of server worker threads, 2 by default.
 runtime_size = 2
-check_pwd = false
+
+# PostgresSQL server TLS options, see `[mysql_options.tls]` section.
+[postgres_options.tls]
+# TLS mode.
+mode = "disable"
+# certificate file path.
+cert_path = ""
+# private key file path.
+key_path = ""
+
+# OpenTSDB protocol options.
+[opentsdb_options]
+# OpenTSDB telnet API server address, "127.0.0.1:4242" by default.
+addr = "127.0.0.1:4242"
+# The number of server worker threads, 2 by default.
+runtime_size = 2
+
+# InfluxDB protocol options.
+[influxdb_options]
+# Whether to enable InfluxDB protocol in HTTP API, true by default.
+enable = true
+
+# Prometheus protocol options.
+[prometheus_options]
+# Whether to enable Prometheus remote write and read in HTTP API, true by default.
+enable = true
+
+# Prom protocol options.
+[prom_options]
+# Prometheus API server address, "127.0.0.1:4004" by default.
+addr = "127.0.0.1:4004"
+
+# WAL options.
+[wal]
+# WAL data directory.
+dir = "/tmp/greptimedb/wal"
+# WAL file size in bytes.
+file_size = "1GB"
+# WAL purge threshold in bytes.
+purge_threshold = "50GB"
+# WAL purge interval in seconds.
+purge_interval = "10m"
+# WAL read batch size.
+read_batch_size = 128
+# Whether to sync log file after every write.
+sync_write = false
+
+# Storage options.
+[storage]
+# Storage type.
+type = "File"
+# Data directory, "/tmp/greptimedb/data" by default.
+data_dir = "/tmp/greptimedb/data/"
+
+# Compaction options.
+[storage.compaction]
+# Max task number that can concurrently run.
+max_inflight_tasks = 4
+# Max files in level 0 to trigger compaction.
+max_files_in_level0 = 8
+# Max task number for SST purge task after compaction.
+max_purge_tasks = 32
+
+# Storage manifest options
+[storage.manifest]
+# Region checkpoint actions margin.
+# Create a checkpoint every <checkpoint_margin> actions.
+checkpoint_margin = 10
+# Region manifest logs and checkpoints gc execution duration
+gc_duration = '30s'
+# Whether to try creating a manifest checkpoint on region opening
+checkpoint_on_startup = false
+
+# Procedure storage options.
+# Uncomment to enable.
+# [procedure.store]
+# # Storage type.
+# type = "File"
+# # Procedure data path.
+# data_dir = "/tmp/greptimedb/procedure/"
+# # Procedure max retry time.
+# max_retry_times = 3
+# # Initial retry delay of procedures, increases exponentially
+# retry_delay = "500ms"
--- a/docker/Dockerfile
+++ b/docker/Dockerfile
@@ -9,7 +9,10 @@ RUN apt-get update && apt-get install -y \
    protobuf-compiler \
    curl \
    build-essential \
-    pkg-config
+    pkg-config \
+    python3 \
+    python3-dev \
+    && pip install pyarrow

 # Install Rust.
 SHELL ["/bin/bash", "-c"]
--- a/docker/aarch64/Dockerfile
+++ b/docker/aarch64/Dockerfile
@@ -0,0 +1,57 @@
+FROM ubuntu:22.04 as builder
+
+ENV LANG en_US.utf8
+WORKDIR /greptimedb
+
+# Install dependencies.
+RUN apt-get update && apt-get install -y \
+    libssl-dev \
+    protobuf-compiler \
+    curl \
+    build-essential \
+    pkg-config \
+    wget
+
+# Install Rust.
+SHELL ["/bin/bash", "-c"]
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- --no-modify-path --default-toolchain none -y
+ENV PATH /root/.cargo/bin/:$PATH
+
+# Install cross platform toolchain
+RUN apt-get -y update && \
+    apt-get -y install g++-aarch64-linux-gnu gcc-aarch64-linux-gnu && \
+    apt-get install binutils-aarch64-linux-gnu
+
+COPY ./docker/aarch64/compile-python.sh ./docker/aarch64/
+RUN chmod +x ./docker/aarch64/compile-python.sh && \
+    ./docker/aarch64/compile-python.sh
+
+COPY ./rust-toolchain.toml .
+# Install rustup target for cross compiling.
+RUN rustup target add aarch64-unknown-linux-gnu
+COPY . .
+# Update dependency, using separate `RUN` to separate cache
+RUN cargo fetch
+
+# This three env var is set in script, so I set it manually in dockerfile.
+ENV LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib/
+ENV LIBRARY_PATH=$LIBRARY_PATH:/usr/local/lib/
+ENV PY_INSTALL_PATH=/greptimedb/python_arm64_build
+
+# Set the environment variable for cross compiling and compile it
+# cross compiled python is `python3` in path, but pyo3 need `python` in path so alias it
+# Build the project in release mode.
+RUN export PYO3_CROSS_LIB_DIR=$PY_INSTALL_PATH/lib && \ 
+    alias python=python3 && \
+    cargo build --target aarch64-unknown-linux-gnu --release -F pyo3_backend
+
+# Exporting the binary to the clean image
+FROM ubuntu:22.04 as base
+
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -y install ca-certificates
+
+WORKDIR /greptime
+COPY --from=builder /greptimedb/target/aarch64-unknown-linux-gnu/release/greptime /greptime/bin/
+ENV PATH /greptime/bin/:$PATH
+
+ENTRYPOINT ["greptime"]
--- a/docker/aarch64/compile-python.sh
+++ b/docker/aarch64/compile-python.sh
@@ -0,0 +1,87 @@
+#!/usr/bin/env bash
+
+set -e
+
+# this script will download Python source code, compile it, and install it to /usr/local/lib
+# then use this python to compile cross-compiled python for aarch64
+ARCH=$1
+PYTHON_VERSION=3.10.10
+PYTHON_SOURCE_DIR=Python-${PYTHON_VERSION}
+PYTHON_INSTALL_PATH_AMD64=${PWD}/python-${PYTHON_VERSION}/amd64
+PYTHON_INSTALL_PATH_AARCH64=${PWD}/python-${PYTHON_VERSION}/aarch64
+
+function download_python_source_code() {
+  wget https://www.python.org/ftp/python/$PYTHON_VERSION/Python-$PYTHON_VERSION.tgz
+  tar -xvf Python-$PYTHON_VERSION.tgz
+}
+
+function compile_for_amd64_platform() {
+  mkdir -p "$PYTHON_INSTALL_PATH_AMD64"
+
+  echo "Compiling for amd64 platform..."
+
+  ./configure \
+    --prefix="$PYTHON_INSTALL_PATH_AMD64" \
+    --enable-shared \
+    ac_cv_pthread_is_default=no ac_cv_pthread=yes ac_cv_cxx_thread=yes \
+    ac_cv_have_long_long_format=yes \
+    --disable-ipv6 ac_cv_file__dev_ptmx=no ac_cv_file__dev_ptc=no
+
+  make
+  make install
+}
+
+# explain Python compile options here a bit:s
+# --enable-shared: enable building a shared Python library (default is no) but we do need it for calling from rust
+# CC, CXX, AR, LD, RANLIB: set the compiler, archiver, linker, and ranlib programs to use
+# build: the machine you are building on, host: the machine you will run the compiled program on
+# --with-system-ffi: build _ctypes module using an installed ffi library, see Doc/library/ctypes.rst, not used in here TODO: could remove
+# ac_cv_pthread_is_default=no ac_cv_pthread=yes ac_cv_cxx_thread=yes:
+# allow cross-compiled python to have -pthread set for CXX, see https://github.com/python/cpython/pull/22525
+# ac_cv_have_long_long_format=yes: target platform supports long long type
+# disable-ipv6: disable ipv6 support, we don't need it in here
+# ac_cv_file__dev_ptmx=no ac_cv_file__dev_ptc=no: disable pty support, we don't need it in here
+function compile_for_aarch64_platform() {
+  export LD_LIBRARY_PATH=$PYTHON_INSTALL_PATH_AMD64/lib:$LD_LIBRARY_PATH
+  export LIBRARY_PATH=$PYTHON_INSTALL_PATH_AMD64/lib:$LIBRARY_PATH
+  export PATH=$PYTHON_INSTALL_PATH_AMD64/bin:$PATH
+
+  mkdir -p "$PYTHON_INSTALL_PATH_AARCH64"
+
+  echo "Compiling for aarch64 platform..."
+  echo "LD_LIBRARY_PATH: $LD_LIBRARY_PATH"
+  echo "LIBRARY_PATH: $LIBRARY_PATH"
+  echo "PATH: $PATH"
+
+  ./configure --build=x86_64-linux-gnu --host=aarch64-linux-gnu \
+    --prefix="$PYTHON_INSTALL_PATH_AARCH64" --enable-optimizations \
+    CC=aarch64-linux-gnu-gcc \
+    CXX=aarch64-linux-gnu-g++ \
+    AR=aarch64-linux-gnu-ar \
+    LD=aarch64-linux-gnu-ld \
+    RANLIB=aarch64-linux-gnu-ranlib \
+    --enable-shared \
+    ac_cv_pthread_is_default=no ac_cv_pthread=yes ac_cv_cxx_thread=yes \
+    ac_cv_have_long_long_format=yes \
+    --disable-ipv6 ac_cv_file__dev_ptmx=no ac_cv_file__dev_ptc=no
+
+  make
+  make altinstall
+}
+
+# Main script starts here.
+download_python_source_code
+
+# Enter the python source code directory.
+cd $PYTHON_SOURCE_DIR || exit 1
+
+# Build local python first, then build cross-compiled python.
+compile_for_amd64_platform
+
+# Clean the build directory.
+make clean && make distclean
+
+# Cross compile python for aarch64.
+if [ "$ARCH" = "aarch64-unknown-linux-gnu" ]; then
+  compile_for_aarch64_platform
+fi
--- a/docker/ci/Dockerfile
+++ b/docker/ci/Dockerfile
@@ -1,6 +1,14 @@
 FROM ubuntu:22.04

-RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -y install ca-certificates
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
+    ca-certificates \
+    python3.10 \
+    python3.10-dev \
+    python3-pip
+
+COPY requirements.txt /etc/greptime/requirements.txt
+
+RUN python3 -m pip install -r /etc/greptime/requirements.txt

 ARG TARGETARCH

--- a/docker/ci/requirements.txt
+++ b/docker/ci/requirements.txt
@@ -0,0 +1,5 @@
+numpy>=1.24.2
+pandas>=1.5.3
+pyarrow>=11.0.0
+requests>=2.28.2
+scipy>=1.10.1
--- a/docs/how-to/how-to-implement-sql-statement.md
+++ b/docs/how-to/how-to-implement-sql-statement.md
@@ -0,0 +1,74 @@
+This document introduces how to implement SQL statements in GreptimeDB.
+
+The execution entry point for SQL statements locates at Frontend Instance. You can see it has
+implemented `SqlQueryHandler`:
+
+```rust
+impl SqlQueryHandler for Instance {
+    type Error = Error;
+
+    async fn do_query(&self, query: &str, query_ctx: QueryContextRef) -> Vec<Result<Output>> {
+        // ...
+    }
+}
+```
+
+Normally, when a SQL query arrives at GreptimeDB, the `do_query` method will be called. After some parsing work, the SQL
+will be feed into `StatementExecutor`:
+
+```rust
+// in Frontend Instance:
+self.statement_executor.execute_sql(stmt, query_ctx).await
+```
+
+That's where we handle our SQL statements. You can just create a new match arm for your statement there, then the
+statement is implemented for both GreptimeDB Standalone and Cluster. You can see how `DESCRIBE TABLE` is implemented as
+an example.
+
+Now, what if the statements should be handled differently for GreptimeDB Standalone and Cluster? You can see there's
+a `SqlStatementExecutor` field in `StatementExecutor`. Each GreptimeDB Standalone and Cluster has its own implementation
+of `SqlStatementExecutor`. If you are going to implement the statements differently in the two mode (
+like `CREATE TABLE`), you have to implement them in their own `SqlStatementExecutor`s.
+
+Summarize as the diagram below:
+
+```text
+                             SQL query                            
+                                |                                
+                                v                                
+                  +---------------------------+                  
+                  | SqlQueryHandler::do_query |                  
+                  +---------------------------+                  
+                                |                                
+                                | SQL parsing                    
+                                v                                
+               +--------------------------------+                
+               | StatementExecutor::execute_sql |                
+               +--------------------------------+                
+                                |                                
+                                | SQL execution                    
+                                v                                
+               +----------------------------------+                
+               | commonly handled statements like |
+               | "plan_exec" for selection or     |
+               +----------------------------------+                
+                       |                |                        
+        For Standalone |                | For Cluster          
+                       v                v                        
+---------------------------+      +---------------------------+ 
+| SqlStatementExecutor impl |      | SqlStatementExecutor impl | 
+| in Datanode Instance      |      | in Frontend DistInstance  | 
+---------------------------+      +---------------------------+ 
+```
+
+Note that some SQL statements can be executed in our QueryEngine, in the form of `LogicalPlan`. You can follow the
+invocation path down to the `QueryEngine` implementation from `StatementExecutor::plan_exec`. For now, there's only
+one `DatafusionQueryEngine` for both GreptimeDB Standalone and Cluster. That lone query engine works for both modes is
+because GreptimeDB read/write data through `Table` trait, and each mode has its own `Table` implementation.
+
+We don't have any bias towards whether statements should be handled in query engine or `StatementExecutor`. You can
+implement one kind of statement in both places. For example, `Insert` with selection is handled in query engine, because
+we can easily do the query part there. However, `Insert` without selection is not, for the cost of parsing statement
+to `LogicalPlan` is not neglectable. So generally if the SQL query is simple enough, you can handle it
+in `StatementExecutor`; otherwise if it is complex or has some part of selection, it should be parsed to `LogicalPlan`
+and handled in query engine.  
--- a/docs/rfcs/2022-12-20-promql-in-rust/rfc.md
+++ b/docs/rfcs/2022-12-20-promql-in-rust/rfc.md
@@ -149,10 +149,10 @@ inputs:
  - title: 'Series Normalize: \noffset = 0'
    operator: prom
    inputs:
-    - title: 'Filter: \ntimetamp > 2022-12-20T10:00:00 && timestamp < 2022-12-21T10:00:00'
+    - title: 'Filter: \ntimestamp > 2022-12-20T10:00:00 && timestamp < 2022-12-21T10:00:00'
      operator: filter
      inputs:
-      - title: 'Table Scan: \ntable = request_duration, timetamp > 2022-12-20T10:00:00 && timestamp < 2022-12-21T10:00:00'
+      - title: 'Table Scan: \ntable = request_duration, timestamp > 2022-12-20T10:00:00 && timestamp < 2022-12-21T10:00:00'
        operator: scan -->

 ![example](example.png)
--- a/docs/rfcs/2023-03-08-region-fault-tolerance.md
+++ b/docs/rfcs/2023-03-08-region-fault-tolerance.md
@@ -0,0 +1,196 @@
+---
+Feature Name: "Fault Tolerance for Region"
+Tracking Issue: https://github.com/GreptimeTeam/greptimedb/issues/1126
+Date: 2023-03-08
+Author: "Luo Fucong <luofucong@greptime.com>"
+---
+
+Fault Tolerance for Region
+----------------------
+
+# Summary
+
+This RFC proposes a method to achieve fault tolerance for regions in GreptimeDB's distributed mode. Or, put it in another way, achieving region high availability("HA") for GreptimeDB cluster. 
+
+In this RFC, we mainly describe two aspects of region HA: how region availability is detected, and what recovery process is need to be taken. We also discuss some alternatives and future work.
+
+When this feature is done, our users could expect a GreptimeDB cluster that can always handle their requests to regions, despite some requests may failed during the region failover. The optimization to reduce the MTTR(Mean Time To Recovery) is not a concern of this RPC, and is left for future work.
+
+# Motivation
+
+Fault tolerance for regions is a critical feature for our clients to use the GreptimeDB cluster confidently. High availability for users to interact with their stored data is a "must have" for any TSDB products, that include our GreptimeDB cluster.
+
+# Details
+
+## Background
+
+Some backgrounds about region in distributed mode:
+
+- A table is logically split into multiple regions. Each region stores a part of non-overlapping table data.
+- Regions are distributed in Datanodes, the mappings are not static, are assigned and governed by Metasrv.
+- In distributed mode, client requests are scoped in regions. To be more specific, when a request that needs to scan multiple regions arrived in Frontend, Frontend splits the request into multiple sub-requests, each of which scans one region only, and submits them to Datanodes that hold corresponding regions. 
+
+In conclusion, as long as regions remain available, and regions could regain availability when failures do occur, the overall region HA could be achieved. With this in mind, let's see how region failures are detected first. 
+
+## Failure Detection
+
+We detect region failures in Metasrv, and do it both passively and actively. Passively means that Metasrv do not fire some "are you healthy" requests to regions. Instead, we carry region healthy information in the heartbeat requests that are submit to Metasrv by Datanodes. 
+
+Datanode already carries its regions stats in the heartbeat request (the non-relevant fields are omitted):
+
+```protobuf
+message HeartbeatRequest {
+  ...
+  // Region stats on this node
+  repeated RegionStat region_stats = 6;
+  ...
+}
+
+message RegionStat {
+  uint64 region_id = 1;
+  TableName table_name = 2;
+  ...
+}
+```
+
+For the sake of simplicity, we don't add another field `bool available = 3` to the `RegionStat` message; instead, if the region were unavailable in the view of the Datanode that contains it, the Datanode just not includes the `RegionStat` of it in the heartbeat request. Or, if the Datanode itself is not unavailable, the heartbeat request is not submitted, effectively the same with not carrying the `RegionStat`.
+
+> The heartbeat interval is now hardcoded to five seconds.
+
+Metasrv gathers the heartbeat requests, extracts the `RegionStat`s, and treat them as region heartbeat. In this way, Metasrv maintains all regions healthy information. If some region's heartbeats were not received in a period of time, Metasrv speculates the region might be unavailable. To make the decision whether a region is failed or not, Metasrv uses a failure detection algorithm called the "[Phi φ Accrual Failure Detection](https://medium.com/@arpitbhayani/phi-%CF%86-accrual-failure-detection-79c21ce53a7a)". Basically, the algorithm calculates a value called "phi" to represent the possibility of a region's unavailability, based on the historical heartbeats' arrived rate. Once the "phi" is above some pre-defined threshold, Metasrv knows the region is failed.
+
+> This algorithm has been widely adopted in some well known products, like Akka and Cassandra.
+
+When Metasrv decides some region is failed from heartbeats, it's not the final decision. Here comes the "actively" detection. Before Metasrv decides to do region failover, it actively invokes the healthy check interface of the Datanode that the failure region resides. Only this healthy check is failed does Metasrv actually start doing failover upon the region.
+
+To conclude, the failure detection pseudo-codes are like this:
+
+```rust
+// in Metasrv:
+fn failure_detection() {
+    loop {
+        // passive detection
+        let failed_regions = all_regions.iter().filter(|r| r.estimated_failure_possibility() > config.phi).collect();
+
+        // find the datanodes that contains the failed regions
+        let datanodes_and_regions = find_region_resides_datanodes(failed_regions);
+
+        // active detection  
+        for (datanode, regions) in datanodes_and_regions {
+            if !datanode.is_healthy(regions) {
+                do_failover(datanode, regions);
+            }
+        }
+
+        sleep(config.detect_interval);
+    }
+}
+```
+
+Some design considerations:
+
+- Why active detecting while we have passively detection? Because it could be happened that the network is singly connectable sometimes (especially in the complex Cloud environment), then the Datanode's heartbeats cannot reach Metasrv, while Metasrv could request Datanode. Active detecting avoid this false positive situation.
+- Why the detection works on region instead of Datanode? Because we might face the possibility that only part of the regions in the Datanode are not available, not ALL regions. Especially the situation that Datanodes are used by multiple tenants. If this is the case, it's better to do failover upon the designated regions instead of the whole regions that reside on the Datanode. All in all, we want a more subtle control over region failover. 
+
+So we detect some regions are not available. How to regain the availability back?
+
+## Region Failover
+
+Region Failover largely relies on remote WAL, aka "[Bunshin](https://github.com/GreptimeTeam/bunshin)". I'm not including any of the details of it in this RFC, let's just assume we already have it.
+
+In general, region failover is fairly simple. Once Metasrv decides to do failover upon some regions, it first chooses one or more Datanodes to hold the failed region. This can be done easily, as the Metasrv already has the whole picture of Datanodes: it knows which Datanode has the minimum regions, what Datanode historically had the lowest CPU usage and IO rate, and how the Datanodes are assigned to tenants, among other information that can all help the Metasrv choose the most suitable Datanodes. Let's call these chosen Datanodes as "candidates".
+
+> The strategy to choose the most suitable candidates required careful design, but it's another RFC.
+
+Then, Metasrv sets the states of these failed regions as "passive". We should add a field to `Region`:
+
+```protobuf
+message Region {
+  uint64 id = 1;
+  string name = 2;
+  Partition partition = 3;
+  
+  message State {
+    Active,
+    Passive,
+  }
+  State state = 4;
+  
+  map<string, string> attrs = 100;
+}
+```
+
+Here `Region` is used in message `RegionRoute`, which indicates how the write request is split among regions. When a region is set as "passive", Frontend knows the write to it should be rejected at the moment (the region read is not blocked, however).
+
+> Making a region "passive" here is effectively blocking the write to it. It's ok in the failover situation, the region is failed anyway. However, when dealing with active maintenance operations, region state requires more refined design. But that's another story. 
+
+Third, Metasrv fires the "close region" requests to the failed Datanodes, and fires the "open region" requests to those candidates. "Close region" requests might be failed due to the unavailability of Datanodes, but that's fine, it's just a best-effort attempt to reduce the chance of any in-flight writes got handled unintentionally after the region is set as "passive". The "open region" requests must have succeeded though. Datanodes open regions from remote WAL.
+
+> Currently the "close region" is undefined in Datanode. It could be a local cache clean up of region data or other resources tidy up.
+
+Finally, when a candidate successfully opens its region, it calls back to Metasrv, indicating it is ready to handle region. "call back" here is backed by its heartbeat to Metasrv. Metasrv updates the region's state to "active", so as to let Frontend lifts the restrictions of region writes (again, the read part of region is untouched).
+
+All the above steps should be managed by remote procedure framework. It's another implementation challenge in the region failover feature. (One is the remote WAL of course.)
+
+A picture is worth a 1000 words:
+
+```text
+                                    +-------------------------+                                        
+                                    | Metasrv detects region  |                                        
+                                    | failure                 |                                        
+                                    +-------------------------+                                        
+                                                 |                                                     
+                                                 v                                                     
+                                    +----------------------------+                                        
+                                    | Metasrv chooses candidates |                                        
+                                    | to hold failed regions     |                                        
+                                    +----------------------------+                                        
+                                                 |                                                     
+                                                 v                                                     
+                                    +-------------------------+       +-------------------------+      
+                                    | Metasrv "passive" the   |------>| Frontend rejects writes |      
+                                    | failed regions          |       | to "passive" regions    |      
+                                    +-------------------------+       +-------------------------+      
+                                                 |                                                     
+                                                 v                                                     
+--------------------------+        +---------------------------+                                        
+| Candidate Datanodes open |<-------| Metasrv fires "close" and |                                        
+| regions from remote WAL  |        | "open" region requests    |                                        
+--------------------------+        +---------------------------+                                        
+             |                                                                                         
+             |                                                                                         
+             |                      +-------------------------+       +-------------------------+      
+             +--------------------->| Metasrv "active" the    |------>| Frontend lifts write    |      
+                                    | failed regions          |       | restriction to regions  |      
+                                    +-------------------------+       +-------------------------+      
+                                                 |                                                     
+                                                 v                                                     
+                                    +-------------------------+                                        
+                                    | Region failover done,   |                                        
+                                    | HA regain               |                                        
+                                    +-------------------------+                                        
+```
+
+# Alternatives
+
+## The "Neon" Way
+
+Remote WAL raises a problem that could harm the write throughput of GreptimeDB cluster: each write request has to do at least two remote call, one is from Frontend to Datanode, and one is from Datanode to remote WAL. What if we do it the "[Neon](https://github.com/neondatabase/neon)" way, making remote WAL sits in between the Frontend and Datanode, couldn't that improve our write throughput? It could, though there're some consistency issues like "read-your-writes" to solve.
+
+However, the main concerns we don't adopt this method are two-fold:
+
+1. Remote WAL is planned to be quorum based, it can be efficiently written;
+2. More importantly, we are planning to make the remote WAL an option that users could choose not to enable it (at the cost of some reliability reduction).
+
+## No WAL, Replication instead
+
+This method replicates region across Datanodes directly, like the common way in shared-nothing database. Were the main region failed, a standby region in the replicate group is elected as new "main" and take the read/write requests. The main concern to this method is the incompatibility to our current architecture and code structure. It requires a major redesign, but gains no significant advantage over the remote WAL method. 
+
+However, the replication does have its own advantage that we can learn from to optimize this failover procedure.
+
+# Future Work
+
+Some optimizations we could take:
+
+- To reduce the MTTR, we could make Metasrv chooses the candidate to each region at normal time. The candidate does some preparation works to reduce the open region time, effectively accelerate the failover procedure.
+- We can adopt the replication method, to the degree that region replicas are used as the fast catch-up candidates. The data difference among replicas is minor, region failover does not need to load or exchange too much data, greatly reduced the region failover time.
--- a/docs/rfcs/2023-03-29-file-external-table/external-table-engine-overview.png
+++ b/docs/rfcs/2023-03-29-file-external-table/external-table-engine-overview.png
--- a/docs/rfcs/2023-03-29-file-external-table/external-table-engine-way-2.png
+++ b/docs/rfcs/2023-03-29-file-external-table/external-table-engine-way-2.png
--- a/docs/rfcs/2023-03-29-file-external-table/rfc.md
+++ b/docs/rfcs/2023-03-29-file-external-table/rfc.md
@@ -0,0 +1,174 @@
+---
+Feature Name: "File external table"
+Tracking Issue: https://github.com/GreptimeTeam/greptimedb/issues/1041
+Date: 2023-03-08
+Author: "Xu Wenkang <wenymedia@gmail.com>"
+---
+
+File external table
+
+---
+
+# Summary
+
+Allows users to perform SQL queries on files
+
+# Motivation
+
+User data may already exist in other storages, i.g., file systems/s3, etc. in CSV, parquet, JSON format, etc. We can provide users the ability to perform SQL queries on these files.
+
+
+# Details
+
+## Overview
+
+The file external table providers users ability to perform SQL queries on these files.
+
+For example, a user has a CSV file on the local file system `/var/data/city.csv`:
+
+```
+Rank , Name , State , 2023 Population , 2020 Census , Annual Change , Density (mi²)
+1 , New York City , New York , 8,992,908 , 8,804,190 , 0.7% , 29,938
+2 , Los Angeles , California , 3,930,586 , 3,898,747 , 0.27% , 8,382
+3 , Chicago , Illinois , 2,761,625 , 2,746,388 , 0.18% , 12,146
+.....
+```
+
+Then user can create a file external table with:
+
+```sql
+CREATE EXTERNAL TABLE city with(location='/var/data/city.csv', format="CSV", field_delimiter = ',', record_delimiter = '\n', skip_header = 1);
+```
+
+Then query the external table with:
+
+```bash
+MySQL> select * from city;
+```
+
+| Rank | Name          | State      | 2023 Population | 2020 Census | Annual Change | Density (mi²) |
+| :--- | :------------ | :--------- | :-------------- | :---------- | :------------ | :------------ |
+| 1    | New York City | New York   | 8,992,908       | 8,804,190   | 0.7%          | 29,938        |
+| 2    | Los Angeles   | California | 3,930,586       | 3,898,747   | 0.27%         | 8,382         |
+| 3    | Chicago       | Illinois   | 2,761,625       | 2,746,388   | 0.18%         | 12,146        |
+
+Drop the external table, if needs with:
+
+```sql
+DROP EXTERNAL TABLE city
+```
+
+
+### Syntax
+
+```
+CREATE EXTERNAL [<database>.]<table_name>
+[
+ (
+    <col_name> <col_type> [NULL | NOT NULL] [COMMENT "<comment>"]
+ )
+]
+[ WITH
+ (
+     LOCATION = 'url'
+   [,FIELD_DELIMITER =  'delimiter' ]
+   [,RECORD_DELIMITER =  'delimiter' ]
+   [,SKIP_HEADER =  '<number>' ]
+   [,FORMAT =  { csv | json | parquet } ]
+   [,PATTERN = '<regex_pattern>' ]
+   [,ENDPOINT = '<uri>' ]
+   [,ACCESS_KEY_ID = '<key_id>' ]
+   [,SECRET_ACCESS_KEY = '<access_key>' ]
+   [,SESSION_TOKEN = '<token>' ]
+   [,REGION = '<region>' ]
+   [,ENABLE_VIRTUAL_HOST_STYLE = '<boolean>']
+   ..
+ )
+]
+```
+
+### Supported File Format
+
+The external file table supports multiple formats; We divide formats into row format and columnar format.
+
+Row formats:
+
+- CSV, JSON
+
+Columnar formats:
+
+- Parquet
+
+Some of these formats support filter pushdown, and others don't. If users use very large files, that format doesn't support pushdown, which might consume a lot of IO for scanning full files and cause a long running query.
+
+### File Table Engine
+
+![overview](external-table-engine-overview.png)
+
+We implement a file table engine that creates an external table by accepting user-specified file paths and treating all records as immutable.
+
+1. File Format Decoder: decode files to the `RecordBatch` stream.
+2. File Table Engine: implement the `TableProvider` trait, store necessary metadata in memory, and provide scan ability.
+
+Our implementation is better for small files. For large files(i.g., a GB-level CSV file), suggests our users import data to the database.
+
+## Drawbacks
+
+- Some formats don't support filter pushdown
+- Hard to support indexing
+
+## Life cycle
+
+### Register a table
+
+1. Write metadata to manifest.
+2. Create the table via file table engine.
+3. Register table to `CatalogProvider` and register table to `SystemCatalog`(persist tables to disk).
+
+### Deregister a table (Drop a table)
+
+1. Fetch the target table info (figure out table engine type).
+2. Deregister the target table in `CatalogProvider` and `SystemCatalog`.
+3. Find the target table engine.
+4. Drop the target table.
+
+### Recover a table when restarting
+
+1. Collect tables name and engine type info.
+2. Find the target tables in different engines.
+3. Open and register tables.
+
+# Alternatives
+
+## Using DataFusion API
+
+We can use datafusion API to register a file table:
+
+```rust
+let ctx = SessionContext::new();
+
+ctx.register_csv("example", "tests/data/example.csv", CsvReadOptions::new()).await?;
+
+// create a plan
+let df = ctx.sql("SELECT a, MIN(b) FROM example WHERE a <= b GROUP BY a LIMIT 100").await?;
+```
+
+### Drawbacks
+
+The DataFusion implements its own `Object Store` abstraction and supports parsing the partitioned directories, which can push down the filter and skips some directories. However, this makes it impossible to use our's `LruCacheLayer`(The parsing of the partitioned directories required paths as input). If we want to manage memory entirely, we should implement our own `TableProvider` or `Table`.
+
+- Impossible to use `CacheLayer`
+
+## Introduce an intermediate representation layer
+
+![overview](external-table-engine-way-2.png)
+
+We convert all files into `parquet` as an intermediate representation. Then we only need to implement a `parquet` file table engine, and we already have a similar one. Also, it supports limited filter pushdown via the `parquet` row group stats.
+
+### Drawbacks
+
+- Computing overhead
+- Storage overhead
+
+
+
--- a/rust-toolchain.toml
+++ b/rust-toolchain.toml
@@ -1,2 +1,2 @@
 [toolchain]
-channel = "nightly-2023-02-14"
+channel = "nightly-2023-02-26"
--- a/scripts/fetch-dashboard-assets.sh
+++ b/scripts/fetch-dashboard-assets.sh
@@ -0,0 +1,39 @@
+#!/usr/bin/env bash
+
+# This script is used to download built dashboard assets from the "GreptimeTeam/dashboard" repository.
+
+set -e
+
+declare -r SCRIPT_DIR=$(cd $(dirname ${0}) >/dev/null 2>&1 && pwd)
+declare -r ROOT_DIR=$(dirname ${SCRIPT_DIR})
+declare -r STATIC_DIR="$ROOT_DIR/src/servers/dashboard"
+
+RELEASE_VERSION="$(cat $STATIC_DIR/VERSION)"
+
+# Download the SHA256 checksum attached to the release. To verify the integrity
+# of the download, this checksum will be used to check the download tar file
+# containing the built dashboard assets.
+curl -Ls https://github.com/GreptimeTeam/dashboard/releases/download/$RELEASE_VERSION/sha256.txt --output sha256.txt
+
+# Download the tar file containing the built dashboard assets.
+curl -L https://github.com/GreptimeTeam/dashboard/releases/download/$RELEASE_VERSION/build.tar.gz --output build.tar.gz
+
+# Verify the checksums match; exit if they don't.
+case "$(uname -s)" in
+    FreeBSD | Darwin)
+        echo "$(cat sha256.txt)" | shasum --algorithm 256 --check \
+            || { echo "Checksums did not match for downloaded dashboard assets!"; exit 1; } ;;
+    Linux)
+        echo "$(cat sha256.txt)" | sha256sum --check -- \
+            || { echo "Checksums did not match for downloaded dashboard assets!"; exit 1; } ;;
+    *)
+        echo "The '$(uname -s)' operating system is not supported as a build host for the dashboard" >&2
+        exit 1
+esac
+
+# Extract the assets and clean up.
+tar -xzf build.tar.gz -C "$STATIC_DIR"
+rm sha256.txt
+rm build.tar.gz
+
+echo "Successfully download dashboard assets to $STATIC_DIR"
--- a/scripts/install.sh
+++ b/scripts/install.sh
@@ -59,5 +59,5 @@ if [ -n "${OS_TYPE}" ] && [ -n "${ARCH_TYPE}" ]; then
        wget "https://github.com/${GITHUB_ORG}/${GITHUB_REPO}/releases/download/${VERSION}/${BIN}-${OS_TYPE}-${ARCH_TYPE}.tgz"
    fi

-    tar xvf ${BIN}-${OS_TYPE}-${ARCH_TYPE}.tgz && rm ${BIN}-${OS_TYPE}-${ARCH_TYPE}.tgz && echo "Run '${BIN} --help' to get started"
+    tar xvf ${BIN}-${OS_TYPE}-${ARCH_TYPE}.tgz && rm ${BIN}-${OS_TYPE}-${ARCH_TYPE}.tgz && echo "Run './${BIN} --help' to get started"
 fi
--- a/src/api/Cargo.toml
+++ b/src/api/Cargo.toml
@@ -10,10 +10,10 @@ common-base = { path = "../common/base" }
 common-error = { path = "../common/error" }
 common-time = { path = "../common/time" }
 datatypes = { path = "../datatypes" }
-greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "3e6349be127b65a8b42a38cda9d527ec423ca77d" }
+greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "0bebe5f69c91cdfbce85cb8f45f9fcd28185261c" }
 prost.workspace = true
 snafu = { version = "0.7", features = ["backtraces"] }
 tonic.workspace = true

 [build-dependencies]
-tonic-build = "0.8"
+tonic-build = "0.9"
--- a/src/api/src/error.rs
+++ b/src/api/src/error.rs
@@ -18,7 +18,7 @@ use common_error::ext::ErrorExt;
 use common_error::prelude::StatusCode;
 use datatypes::prelude::ConcreteDataType;
 use snafu::prelude::*;
-use snafu::{Backtrace, ErrorCompat};
+use snafu::Location;

 pub type Result<T> = std::result::Result<T, Error>;

@@ -26,12 +26,12 @@ pub type Result<T> = std::result::Result<T, Error>;
 #[snafu(visibility(pub))]
 pub enum Error {
    #[snafu(display("Unknown proto column datatype: {}", datatype))]
-    UnknownColumnDataType { datatype: i32, backtrace: Backtrace },
+    UnknownColumnDataType { datatype: i32, location: Location },

    #[snafu(display("Failed to create column datatype from {:?}", from))]
    IntoColumnDataType {
        from: ConcreteDataType,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display(
@@ -66,9 +66,6 @@ impl ErrorExt for Error {
            | Error::InvalidColumnDefaultConstraint { source, .. } => source.status_code(),
        }
    }
-    fn backtrace_opt(&self) -> Option<&Backtrace> {
-        ErrorCompat::backtrace(self)
-    }

    fn as_any(&self) -> &dyn Any {
        self
--- a/src/api/src/helper.rs
+++ b/src/api/src/helper.rs
@@ -97,7 +97,9 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
                TimestampType::Microsecond(_) => ColumnDataType::TimestampMicrosecond,
                TimestampType::Nanosecond(_) => ColumnDataType::TimestampNanosecond,
            },
-            ConcreteDataType::Null(_) | ConcreteDataType::List(_) => {
+            ConcreteDataType::Null(_)
+            | ConcreteDataType::List(_)
+            | ConcreteDataType::Dictionary(_) => {
                return error::IntoColumnDataTypeSnafu { from: datatype }.fail()
            }
        });
--- a/src/catalog/Cargo.toml
+++ b/src/catalog/Cargo.toml
@@ -7,6 +7,7 @@ license.workspace = true
 [dependencies]
 api = { path = "../api" }
 arc-swap = "1.0"
+arrow-schema.workspace = true
 async-stream.workspace = true
 async-trait = "0.1"
 backoff = { version = "0.4", features = ["tokio"] }
@@ -18,25 +19,28 @@ common-recordbatch = { path = "../common/recordbatch" }
 common-runtime = { path = "../common/runtime" }
 common-telemetry = { path = "../common/telemetry" }
 common-time = { path = "../common/time" }
+dashmap = "5.4"
 datafusion.workspace = true
 datatypes = { path = "../datatypes" }
 futures = "0.3"
 futures-util.workspace = true
 lazy_static = "1.4"
 meta-client = { path = "../meta-client" }
+parking_lot = "0.12"
 regex = "1.6"
 serde = "1.0"
 serde_json = "1.0"
+session = { path = "../session" }
 snafu = { version = "0.7", features = ["backtraces"] }
 storage = { path = "../storage" }
 table = { path = "../table" }
 tokio.workspace = true

 [dev-dependencies]
+common-test-util = { path = "../common/test-util" }
 chrono.workspace = true
 log-store = { path = "../log-store" }
 mito = { path = "../mito", features = ["test"] }
 object-store = { path = "../object-store" }
 storage = { path = "../storage" }
-tempdir = "0.3"
 tokio.workspace = true
--- a/src/catalog/src/datafusion.rs
+++ b/src/catalog/src/datafusion.rs
@@ -12,4 +12,4 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-//! GreptimeDB builtin functions
+pub mod catalog_adapter;
--- a/src/catalog/src/datafusion/catalog_adapter.rs
+++ b/src/catalog/src/datafusion/catalog_adapter.rs
@@ -17,10 +17,7 @@
 use std::any::Any;
 use std::sync::Arc;

-use catalog::error::{self as catalog_error, Error};
-use catalog::{
-    CatalogListRef, CatalogProvider, CatalogProviderRef, SchemaProvider, SchemaProviderRef,
-};
+use async_trait::async_trait;
 use common_error::prelude::BoxedError;
 use datafusion::catalog::catalog::{
    CatalogList as DfCatalogList, CatalogProvider as DfCatalogProvider,
@@ -32,7 +29,10 @@ use snafu::ResultExt;
 use table::table::adapter::{DfTableProviderAdapter, TableAdapter};
 use table::TableRef;

-use crate::datafusion::error;
+use crate::error::{self, Result, SchemaProviderOperationSnafu};
+use crate::{
+    CatalogListRef, CatalogProvider, CatalogProviderRef, SchemaProvider, SchemaProviderRef,
+};

 pub struct DfCatalogListAdapter {
    catalog_list: CatalogListRef,
@@ -88,7 +88,7 @@ impl CatalogProvider for CatalogProviderAdapter {
        self
    }

-    fn schema_names(&self) -> catalog::error::Result<Vec<String>> {
+    fn schema_names(&self) -> Result<Vec<String>> {
        Ok(self.df_catalog_provider.schema_names())
    }

@@ -96,11 +96,11 @@ impl CatalogProvider for CatalogProviderAdapter {
        &self,
        _name: String,
        _schema: SchemaProviderRef,
-    ) -> catalog::error::Result<Option<SchemaProviderRef>> {
+    ) -> Result<Option<SchemaProviderRef>> {
        todo!("register_schema is not supported in Datafusion catalog provider")
    }

-    fn schema(&self, name: &str) -> catalog::error::Result<Option<Arc<dyn SchemaProvider>>> {
+    fn schema(&self, name: &str) -> Result<Option<Arc<dyn SchemaProvider>>> {
        Ok(self
            .df_catalog_provider
            .schema(name)
@@ -137,6 +137,7 @@ struct DfSchemaProviderAdapter {
    schema_provider: Arc<dyn SchemaProvider>,
 }

+#[async_trait]
 impl DfSchemaProvider for DfSchemaProviderAdapter {
    fn as_any(&self) -> &dyn Any {
        self
@@ -148,9 +149,10 @@ impl DfSchemaProvider for DfSchemaProviderAdapter {
            .expect("datafusion does not accept fallible catalog access")
    }

-    fn table(&self, name: &str) -> Option<Arc<dyn DfTableProvider>> {
+    async fn table(&self, name: &str) -> Option<Arc<dyn DfTableProvider>> {
        self.schema_provider
            .table(name)
+            .await
            .expect("datafusion does not accept fallible catalog access")
            .map(|table| Arc::new(DfTableProviderAdapter::new(table)) as _)
    }
@@ -186,18 +188,20 @@ struct SchemaProviderAdapter {
    df_schema_provider: Arc<dyn DfSchemaProvider>,
 }

+#[async_trait]
 impl SchemaProvider for SchemaProviderAdapter {
    fn as_any(&self) -> &dyn Any {
        self
    }

    /// Retrieves the list of available table names in this schema.
-    fn table_names(&self) -> Result<Vec<String>, Error> {
+    fn table_names(&self) -> Result<Vec<String>> {
        Ok(self.df_schema_provider.table_names())
    }

-    fn table(&self, name: &str) -> Result<Option<TableRef>, Error> {
-        let table = self.df_schema_provider.table(name).map(|table_provider| {
+    async fn table(&self, name: &str) -> Result<Option<TableRef>> {
+        let table = self.df_schema_provider.table(name).await;
+        let table = table.map(|table_provider| {
            match table_provider
                .as_any()
                .downcast_ref::<DfTableProviderAdapter>()
@@ -214,11 +218,7 @@ impl SchemaProvider for SchemaProviderAdapter {
        Ok(table)
    }

-    fn register_table(
-        &self,
-        name: String,
-        table: TableRef,
-    ) -> catalog::error::Result<Option<TableRef>> {
+    fn register_table(&self, name: String, table: TableRef) -> Result<Option<TableRef>> {
        let table_provider = Arc::new(DfTableProviderAdapter::new(table.clone()));
        Ok(self
            .df_schema_provider
@@ -227,43 +227,43 @@ impl SchemaProvider for SchemaProviderAdapter {
                msg: "Fail to register table to datafusion",
            })
            .map_err(BoxedError::new)
-            .context(catalog_error::SchemaProviderOperationSnafu)?
+            .context(SchemaProviderOperationSnafu)?
            .map(|_| table))
    }

-    fn rename_table(&self, _name: &str, _new_name: String) -> catalog_error::Result<TableRef> {
+    fn rename_table(&self, _name: &str, _new_name: String) -> Result<TableRef> {
        todo!()
    }

-    fn deregister_table(&self, name: &str) -> catalog::error::Result<Option<TableRef>> {
+    fn deregister_table(&self, name: &str) -> Result<Option<TableRef>> {
        self.df_schema_provider
            .deregister_table(name)
            .context(error::DatafusionSnafu {
                msg: "Fail to deregister table from datafusion",
            })
            .map_err(BoxedError::new)
-            .context(catalog_error::SchemaProviderOperationSnafu)?
+            .context(SchemaProviderOperationSnafu)?
            .map(|table| {
                let adapter = TableAdapter::new(table)
                    .context(error::TableSchemaMismatchSnafu)
                    .map_err(BoxedError::new)
-                    .context(catalog_error::SchemaProviderOperationSnafu)?;
+                    .context(SchemaProviderOperationSnafu)?;
                Ok(Arc::new(adapter) as _)
            })
            .transpose()
    }

-    fn table_exist(&self, name: &str) -> Result<bool, Error> {
+    fn table_exist(&self, name: &str) -> Result<bool> {
        Ok(self.df_schema_provider.table_exist(name))
    }
 }

 #[cfg(test)]
 mod tests {
-    use catalog::local::{new_memory_catalog_list, MemoryCatalogProvider, MemorySchemaProvider};
    use table::table::numbers::NumbersTable;

    use super::*;
+    use crate::local::{new_memory_catalog_list, MemoryCatalogProvider, MemorySchemaProvider};

    #[test]
    #[should_panic]
@@ -282,8 +282,8 @@ mod tests {
            .unwrap();
    }

-    #[test]
-    pub fn test_register_table() {
+    #[tokio::test]
+    async fn test_register_table() {
        let adapter = DfSchemaProviderAdapter {
            schema_provider: Arc::new(MemorySchemaProvider::new()),
        };
@@ -296,7 +296,7 @@ mod tests {
                ))),
            )
            .unwrap();
-        adapter.table("test_table").unwrap();
+        adapter.table("test_table").await.unwrap();
    }

    #[test]
--- a/src/catalog/src/error.rs
+++ b/src/catalog/src/error.rs
@@ -19,7 +19,7 @@ use common_error::ext::{BoxedError, ErrorExt};
 use common_error::prelude::{Snafu, StatusCode};
 use datafusion::error::DataFusionError;
 use datatypes::prelude::ConcreteDataType;
-use snafu::{Backtrace, ErrorCompat};
+use snafu::Location;

 use crate::DeregisterTableRequest;

@@ -50,7 +50,7 @@ pub enum Error {
    },

    #[snafu(display("System catalog is not valid: {}", msg))]
-    SystemCatalog { msg: String, backtrace: Backtrace },
+    SystemCatalog { msg: String, location: Location },

    #[snafu(display(
        "System catalog table type mismatch, expected: binary, found: {:?}",
@@ -58,61 +58,68 @@ pub enum Error {
    ))]
    SystemCatalogTypeMismatch {
        data_type: ConcreteDataType,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Invalid system catalog entry type: {:?}", entry_type))]
    InvalidEntryType {
        entry_type: Option<u8>,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Invalid system catalog key: {:?}", key))]
    InvalidKey {
        key: Option<String>,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Catalog value is not present"))]
-    EmptyValue { backtrace: Backtrace },
+    EmptyValue { location: Location },

    #[snafu(display("Failed to deserialize value, source: {}", source))]
    ValueDeserialize {
        source: serde_json::error::Error,
-        backtrace: Backtrace,
+        location: Location,
+    },
+
+    #[snafu(display("Table engine not found: {}, source: {}", engine_name, source))]
+    TableEngineNotFound {
+        engine_name: String,
+        #[snafu(backtrace)]
+        source: table::error::Error,
    },

    #[snafu(display("Cannot find catalog by name: {}", catalog_name))]
    CatalogNotFound {
        catalog_name: String,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Cannot find schema {} in catalog {}", schema, catalog))]
    SchemaNotFound {
        catalog: String,
        schema: String,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Table `{}` already exists", table))]
-    TableExists { table: String, backtrace: Backtrace },
+    TableExists { table: String, location: Location },

    #[snafu(display("Table `{}` not exist", table))]
-    TableNotExist { table: String, backtrace: Backtrace },
+    TableNotExist { table: String, location: Location },

    #[snafu(display("Schema {} already exists", schema))]
-    SchemaExists {
-        schema: String,
-        backtrace: Backtrace,
-    },
+    SchemaExists { schema: String, location: Location },

    #[snafu(display("Operation {} not implemented yet", operation))]
    Unimplemented {
        operation: String,
-        backtrace: Backtrace,
+        location: Location,
    },

+    #[snafu(display("Operation {} not supported", op))]
+    NotSupported { op: String, location: Location },
+
    #[snafu(display("Failed to open table, table info: {}, source: {}", table_info, source))]
    OpenTable {
        table_info: String,
@@ -123,7 +130,7 @@ pub enum Error {
    #[snafu(display("Table not found while opening table, table info: {}", table_info))]
    TableNotFound {
        table_info: String,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Failed to read system catalog table records"))]
@@ -132,6 +139,12 @@ pub enum Error {
        source: common_recordbatch::error::Error,
    },

+    #[snafu(display("Failed to create recordbatch, source: {}", source))]
+    CreateRecordBatch {
+        #[snafu(backtrace)]
+        source: common_recordbatch::error::Error,
+    },
+
    #[snafu(display(
        "Failed to insert table creation record to system catalog, source: {}",
        source
@@ -153,7 +166,7 @@ pub enum Error {
    },

    #[snafu(display("Illegal catalog manager state: {}", msg))]
-    IllegalManagerState { backtrace: Backtrace, msg: String },
+    IllegalManagerState { location: Location, msg: String },

    #[snafu(display("Failed to scan system catalog table, source: {}", source))]
    SystemCatalogTableScan {
@@ -201,6 +214,40 @@ pub enum Error {
        #[snafu(backtrace)]
        source: common_catalog::error::Error,
    },
+
+    #[snafu(display("Illegal access to catalog: {} and schema: {}", catalog, schema))]
+    QueryAccessDenied { catalog: String, schema: String },
+
+    #[snafu(display(
+        "Failed to get region stats, catalog: {}, schema: {}, table: {}, source: {}",
+        catalog,
+        schema,
+        table,
+        source
+    ))]
+    RegionStats {
+        catalog: String,
+        schema: String,
+        table: String,
+        #[snafu(backtrace)]
+        source: table::error::Error,
+    },
+
+    #[snafu(display("Invalid system table definition: {err_msg}"))]
+    InvalidSystemTableDef { err_msg: String, location: Location },
+
+    #[snafu(display("{}: {}", msg, source))]
+    Datafusion {
+        msg: String,
+        source: DataFusionError,
+        location: Location,
+    },
+
+    #[snafu(display("Table schema mismatch, source: {}", source))]
+    TableSchemaMismatch {
+        #[snafu(backtrace)]
+        source: table::error::Error,
+    },
 }

 pub type Result<T> = std::result::Result<T, Error>;
@@ -213,7 +260,8 @@ impl ErrorExt for Error {
            | Error::TableNotFound { .. }
            | Error::IllegalManagerState { .. }
            | Error::CatalogNotFound { .. }
-            | Error::InvalidEntryType { .. } => StatusCode::Unexpected,
+            | Error::InvalidEntryType { .. }
+            | Error::InvalidSystemTableDef { .. } => StatusCode::Unexpected,

            Error::SystemCatalog { .. }
            | Error::EmptyValue { .. }
@@ -221,21 +269,27 @@ impl ErrorExt for Error {

            Error::SystemCatalogTypeMismatch { .. } => StatusCode::Internal,

-            Error::ReadSystemCatalog { source, .. } => source.status_code(),
+            Error::ReadSystemCatalog { source, .. } | Error::CreateRecordBatch { source } => {
+                source.status_code()
+            }
            Error::InvalidCatalogValue { source, .. } | Error::CatalogEntrySerde { source } => {
                source.status_code()
            }

            Error::TableExists { .. } => StatusCode::TableAlreadyExists,
            Error::TableNotExist { .. } => StatusCode::TableNotFound,
-            Error::SchemaExists { .. } => StatusCode::InvalidArguments,
+            Error::SchemaExists { .. } | Error::TableEngineNotFound { .. } => {
+                StatusCode::InvalidArguments
+            }

            Error::OpenSystemCatalog { source, .. }
            | Error::CreateSystemCatalog { source, .. }
            | Error::InsertCatalogRecord { source, .. }
            | Error::OpenTable { source, .. }
            | Error::CreateTable { source, .. }
-            | Error::DeregisterTable { source, .. } => source.status_code(),
+            | Error::DeregisterTable { source, .. }
+            | Error::RegionStats { source, .. }
+            | Error::TableSchemaMismatch { source } => source.status_code(),

            Error::MetaSrv { source, .. } => source.status_code(),
            Error::SystemCatalogTableScan { source } => source.status_code(),
@@ -245,14 +299,12 @@ impl ErrorExt for Error {
                source.status_code()
            }

-            Error::Unimplemented { .. } => StatusCode::Unsupported,
+            Error::Unimplemented { .. } | Error::NotSupported { .. } => StatusCode::Unsupported,
+            Error::QueryAccessDenied { .. } => StatusCode::AccessDenied,
+            Error::Datafusion { .. } => StatusCode::EngineExecuteQuery,
        }
    }

-    fn backtrace_opt(&self) -> Option<&Backtrace> {
-        ErrorCompat::backtrace(self)
-    }
-
    fn as_any(&self) -> &dyn Any {
        self
    }
@@ -276,7 +328,7 @@ mod tests {
            StatusCode::TableAlreadyExists,
            Error::TableExists {
                table: "some_table".to_string(),
-                backtrace: Backtrace::generate(),
+                location: Location::generate(),
            }
            .status_code()
        );
@@ -290,7 +342,7 @@ mod tests {
            StatusCode::StorageUnavailable,
            Error::SystemCatalog {
                msg: "".to_string(),
-                backtrace: Backtrace::generate(),
+                location: Location::generate(),
            }
            .status_code()
        );
@@ -299,7 +351,7 @@ mod tests {
            StatusCode::Internal,
            Error::SystemCatalogTypeMismatch {
                data_type: ConcreteDataType::binary_datatype(),
-                backtrace: Backtrace::generate(),
+                location: Location::generate(),
            }
            .status_code()
        );
@@ -313,7 +365,7 @@ mod tests {
    pub fn test_errors_to_datafusion_error() {
        let e: DataFusionError = Error::TableExists {
            table: "test_table".to_string(),
-            backtrace: Backtrace::generate(),
+            location: Location::generate(),
        }
        .into();
        match e {
--- a/src/catalog/src/helper.rs
+++ b/src/catalog/src/helper.rs
@@ -24,10 +24,10 @@ use serde::{Deserialize, Serialize, Serializer};
 use snafu::{ensure, OptionExt, ResultExt};
 use table::metadata::{RawTableInfo, TableId, TableVersion};

-const CATALOG_KEY_PREFIX: &str = "__c";
-const SCHEMA_KEY_PREFIX: &str = "__s";
-const TABLE_GLOBAL_KEY_PREFIX: &str = "__tg";
-const TABLE_REGIONAL_KEY_PREFIX: &str = "__tr";
+pub const CATALOG_KEY_PREFIX: &str = "__c";
+pub const SCHEMA_KEY_PREFIX: &str = "__s";
+pub const TABLE_GLOBAL_KEY_PREFIX: &str = "__tg";
+pub const TABLE_REGIONAL_KEY_PREFIX: &str = "__tr";

 const ALPHANUMERICS_NAME_PATTERN: &str = "[a-zA-Z_][a-zA-Z0-9_]*";

@@ -370,4 +370,10 @@ mod tests {
        let deserialized = TableGlobalValue::parse(serialized).unwrap();
        assert_eq!(value, deserialized);
    }
+
+    #[test]
+    fn test_table_global_value_compatibility() {
+        let s = r#"{"node_id":1,"regions_id_map":{"1":[0]},"table_info":{"ident":{"table_id":1098,"version":1},"name":"container_cpu_limit","desc":"Created on insertion","catalog_name":"greptime","schema_name":"dd","meta":{"schema":{"column_schemas":[{"name":"container_id","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"container_name","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"docker_image","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"host","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"image_name","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"image_tag","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"interval","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"runtime","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"short_image","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"type","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"dd_value","data_type":{"Float64":{}},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}},{"name":"ts","data_type":{"Timestamp":{"Millisecond":null}},"is_nullable":false,"is_time_index":true,"default_constraint":null,"metadata":{"greptime:time_index":"true"}},{"name":"git.repository_url","data_type":{"String":null},"is_nullable":true,"is_time_index":false,"default_constraint":null,"metadata":{}}],"timestamp_index":11,"version":1},"primary_key_indices":[0,1,2,3,4,5,6,7,8,9,12],"value_indices":[10,11],"engine":"mito","next_column_id":12,"region_numbers":[],"engine_options":{},"options":{},"created_on":"1970-01-01T00:00:00Z"},"table_type":"Base"}}"#;
+        TableGlobalValue::parse(s).unwrap();
+    }
 }
--- a/src/catalog/src/information_schema.rs
+++ b/src/catalog/src/information_schema.rs
@@ -0,0 +1,80 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+mod tables;
+
+use std::any::Any;
+use std::sync::Arc;
+
+use async_trait::async_trait;
+use datafusion::datasource::streaming::{PartitionStream, StreamingTable};
+use snafu::ResultExt;
+use table::table::adapter::TableAdapter;
+use table::TableRef;
+
+use crate::error::{DatafusionSnafu, Result, TableSchemaMismatchSnafu};
+use crate::information_schema::tables::InformationSchemaTables;
+use crate::{CatalogProviderRef, SchemaProvider};
+
+const TABLES: &str = "tables";
+
+pub(crate) struct InformationSchemaProvider {
+    catalog_name: String,
+    catalog_provider: CatalogProviderRef,
+}
+
+impl InformationSchemaProvider {
+    pub(crate) fn new(catalog_name: String, catalog_provider: CatalogProviderRef) -> Self {
+        Self {
+            catalog_name,
+            catalog_provider,
+        }
+    }
+}
+
+#[async_trait]
+impl SchemaProvider for InformationSchemaProvider {
+    fn as_any(&self) -> &dyn Any {
+        self
+    }
+
+    fn table_names(&self) -> Result<Vec<String>> {
+        Ok(vec![TABLES.to_string()])
+    }
+
+    async fn table(&self, name: &str) -> Result<Option<TableRef>> {
+        let table = if name.eq_ignore_ascii_case(TABLES) {
+            Arc::new(InformationSchemaTables::new(
+                self.catalog_name.clone(),
+                self.catalog_provider.clone(),
+            ))
+        } else {
+            return Ok(None);
+        };
+
+        let table = Arc::new(
+            StreamingTable::try_new(table.schema().clone(), vec![table]).with_context(|_| {
+                DatafusionSnafu {
+                    msg: format!("Failed to get InformationSchema table '{name}'"),
+                }
+            })?,
+        );
+        let table = TableAdapter::new(table).context(TableSchemaMismatchSnafu)?;
+        Ok(Some(Arc::new(table)))
+    }
+
+    fn table_exist(&self, name: &str) -> Result<bool> {
+        Ok(matches!(name.to_ascii_lowercase().as_str(), TABLES))
+    }
+}
--- a/src/catalog/src/information_schema/tables.rs
+++ b/src/catalog/src/information_schema/tables.rs
@@ -0,0 +1,165 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::sync::Arc;
+
+use arrow_schema::SchemaRef as ArrowSchemaRef;
+use common_catalog::consts::INFORMATION_SCHEMA_NAME;
+use common_query::physical_plan::TaskContext;
+use common_recordbatch::RecordBatch;
+use datafusion::datasource::streaming::PartitionStream as DfPartitionStream;
+use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
+use datafusion::physical_plan::SendableRecordBatchStream as DfSendableRecordBatchStream;
+use datatypes::prelude::{ConcreteDataType, ScalarVectorBuilder, VectorRef};
+use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
+use datatypes::vectors::StringVectorBuilder;
+use snafu::ResultExt;
+use table::metadata::TableType;
+
+use crate::error::{CreateRecordBatchSnafu, Result};
+use crate::information_schema::TABLES;
+use crate::CatalogProviderRef;
+
+pub(super) struct InformationSchemaTables {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_provider: CatalogProviderRef,
+}
+
+impl InformationSchemaTables {
+    pub(super) fn new(catalog_name: String, catalog_provider: CatalogProviderRef) -> Self {
+        let schema = Arc::new(Schema::new(vec![
+            ColumnSchema::new("table_catalog", ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new("table_schema", ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new("table_name", ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new("table_type", ConcreteDataType::string_datatype(), false),
+        ]));
+        Self {
+            schema,
+            catalog_name,
+            catalog_provider,
+        }
+    }
+
+    fn builder(&self) -> InformationSchemaTablesBuilder {
+        InformationSchemaTablesBuilder::new(
+            self.schema.clone(),
+            self.catalog_name.clone(),
+            self.catalog_provider.clone(),
+        )
+    }
+}
+
+/// Builds the `information_schema.TABLE` table row by row
+///
+/// Columns are based on <https://www.postgresql.org/docs/current/infoschema-columns.html>
+struct InformationSchemaTablesBuilder {
+    schema: SchemaRef,
+    catalog_name: String,
+    catalog_provider: CatalogProviderRef,
+
+    catalog_names: StringVectorBuilder,
+    schema_names: StringVectorBuilder,
+    table_names: StringVectorBuilder,
+    table_types: StringVectorBuilder,
+}
+
+impl InformationSchemaTablesBuilder {
+    fn new(schema: SchemaRef, catalog_name: String, catalog_provider: CatalogProviderRef) -> Self {
+        Self {
+            schema,
+            catalog_name,
+            catalog_provider,
+            catalog_names: StringVectorBuilder::with_capacity(42),
+            schema_names: StringVectorBuilder::with_capacity(42),
+            table_names: StringVectorBuilder::with_capacity(42),
+            table_types: StringVectorBuilder::with_capacity(42),
+        }
+    }
+
+    /// Construct the `information_schema.tables` virtual table
+    async fn make_tables(&mut self) -> Result<RecordBatch> {
+        let catalog_name = self.catalog_name.clone();
+
+        for schema_name in self.catalog_provider.schema_names()? {
+            if schema_name == INFORMATION_SCHEMA_NAME {
+                continue;
+            }
+
+            let Some(schema) = self.catalog_provider.schema(&schema_name)? else { continue };
+            for table_name in schema.table_names()? {
+                let Some(table) = schema.table(&table_name).await? else { continue };
+                self.add_table(&catalog_name, &schema_name, &table_name, table.table_type());
+            }
+        }
+
+        // Add a final list for the information schema tables themselves
+        self.add_table(
+            &catalog_name,
+            INFORMATION_SCHEMA_NAME,
+            TABLES,
+            TableType::View,
+        );
+
+        self.finish()
+    }
+
+    fn add_table(
+        &mut self,
+        catalog_name: &str,
+        schema_name: &str,
+        table_name: &str,
+        table_type: TableType,
+    ) {
+        self.catalog_names.push(Some(catalog_name));
+        self.schema_names.push(Some(schema_name));
+        self.table_names.push(Some(table_name));
+        self.table_types.push(Some(match table_type {
+            TableType::Base => "BASE TABLE",
+            TableType::View => "VIEW",
+            TableType::Temporary => "LOCAL TEMPORARY",
+        }));
+    }
+
+    fn finish(&mut self) -> Result<RecordBatch> {
+        let columns: Vec<VectorRef> = vec![
+            Arc::new(self.catalog_names.finish()),
+            Arc::new(self.schema_names.finish()),
+            Arc::new(self.table_names.finish()),
+            Arc::new(self.table_types.finish()),
+        ];
+        RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
+    }
+}
+
+impl DfPartitionStream for InformationSchemaTables {
+    fn schema(&self) -> &ArrowSchemaRef {
+        self.schema.arrow_schema()
+    }
+
+    fn execute(&self, _: Arc<TaskContext>) -> DfSendableRecordBatchStream {
+        let schema = self.schema().clone();
+        let mut builder = self.builder();
+        Box::pin(DfRecordBatchStreamAdapter::new(
+            schema,
+            futures::stream::once(async move {
+                builder
+                    .make_tables()
+                    .await
+                    .map(|x| x.into_df_record_batch())
+                    .map_err(Into::into)
+            }),
+        ))
+    }
+}
--- a/src/catalog/src/lib.rs
+++ b/src/catalog/src/lib.rs
@@ -18,8 +18,9 @@ use std::any::Any;
 use std::fmt::{Debug, Formatter};
 use std::sync::Arc;

-use common_telemetry::info;
-use snafu::{OptionExt, ResultExt};
+use api::v1::meta::{RegionStat, TableName};
+use common_telemetry::{info, warn};
+use snafu::ResultExt;
 use table::engine::{EngineContext, TableEngineRef};
 use table::metadata::TableId;
 use table::requests::CreateTableRequest;
@@ -28,12 +29,15 @@ use table::TableRef;
 use crate::error::{CreateTableSnafu, Result};
 pub use crate::schema::{SchemaProvider, SchemaProviderRef};

+pub mod datafusion;
 pub mod error;
 pub mod helper;
+pub(crate) mod information_schema;
 pub mod local;
 pub mod remote;
 pub mod schema;
 pub mod system;
+pub mod table_source;
 pub mod tables;

 /// Represent a list of named catalogs
@@ -107,7 +111,12 @@ pub trait CatalogManager: CatalogList {
    fn schema(&self, catalog: &str, schema: &str) -> Result<Option<SchemaProviderRef>>;

    /// Returns the table by catalog, schema and table name.
-    fn table(&self, catalog: &str, schema: &str, table_name: &str) -> Result<Option<TableRef>>;
+    async fn table(
+        &self,
+        catalog: &str,
+        schema: &str,
+        table_name: &str,
+    ) -> Result<Option<TableRef>>;
 }

 pub type CatalogManagerRef = Arc<dyn CatalogManager>;
@@ -186,7 +195,8 @@ pub(crate) async fn handle_system_table_request<'a, M: CatalogManager>(
        let table_name = &req.create_table_request.table_name;
        let table_id = req.create_table_request.id;

-        let table = if let Some(table) = manager.table(catalog_name, schema_name, table_name)? {
+        let table = manager.table(catalog_name, schema_name, table_name).await?;
+        let table = if let Some(table) = table {
            table
        } else {
            let table = engine
@@ -218,37 +228,52 @@ pub(crate) async fn handle_system_table_request<'a, M: CatalogManager>(
    Ok(())
 }

-/// The number of regions in the datanode node.
-pub fn region_number(catalog_manager: &CatalogManagerRef) -> Result<u64> {
+/// The stat of regions in the datanode node.
+/// The number of regions can be got from len of vec.
+///
+/// Ignores any errors occurred during iterating regions. The intention of this method is to
+/// collect region stats that will be carried in Datanode's heartbeat to Metasrv, so it's a
+/// "try our best" job.
+pub async fn datanode_stat(catalog_manager: &CatalogManagerRef) -> (u64, Vec<RegionStat>) {
    let mut region_number: u64 = 0;
+    let mut region_stats = Vec::new();

-    for catalog_name in catalog_manager.catalog_names()? {
-        let catalog =
-            catalog_manager
-                .catalog(&catalog_name)?
-                .context(error::CatalogNotFoundSnafu {
-                    catalog_name: &catalog_name,
-                })?;
+    let Ok(catalog_names) = catalog_manager.catalog_names() else { return (region_number, region_stats) };
+    for catalog_name in catalog_names {
+        let Ok(Some(catalog)) = catalog_manager.catalog(&catalog_name) else { continue };

-        for schema_name in catalog.schema_names()? {
-            let schema = catalog
-                .schema(&schema_name)?
-                .context(error::SchemaNotFoundSnafu {
-                    catalog: &catalog_name,
-                    schema: &schema_name,
-                })?;
+        let Ok(schema_names) = catalog.schema_names() else { continue };
+        for schema_name in schema_names {
+            let Ok(Some(schema)) = catalog.schema(&schema_name) else { continue };

-            for table_name in schema.table_names()? {
-                let table = schema
-                    .table(&table_name)?
-                    .context(error::TableNotFoundSnafu {
-                        table_info: &table_name,
-                    })?;
+            let Ok(table_names) = schema.table_names() else { continue };
+            for table_name in table_names {
+                let Ok(Some(table)) = schema.table(&table_name).await else { continue };

                let region_numbers = &table.table_info().meta.region_numbers;
                region_number += region_numbers.len() as u64;
+
+                match table.region_stats() {
+                    Ok(stats) => {
+                        let stats = stats.into_iter().map(|stat| RegionStat {
+                            region_id: stat.region_id,
+                            table_name: Some(TableName {
+                                catalog_name: catalog_name.clone(),
+                                schema_name: schema_name.clone(),
+                                table_name: table_name.clone(),
+                            }),
+                            approximate_bytes: stat.disk_usage_bytes as i64,
+                            ..Default::default()
+                        });
+
+                        region_stats.extend(stats);
+                    }
+                    Err(e) => {
+                        warn!("Failed to get region status, err: {:?}", e);
+                    }
+                };
            }
        }
    }
-    Ok(region_number)
+    (region_number, region_stats)
 }
--- a/src/catalog/src/local/manager.rs
+++ b/src/catalog/src/local/manager.rs
@@ -18,7 +18,7 @@ use std::sync::Arc;

 use common_catalog::consts::{
    DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, MIN_USER_TABLE_ID,
-    SYSTEM_CATALOG_NAME, SYSTEM_CATALOG_TABLE_NAME,
+    MITO_ENGINE, SYSTEM_CATALOG_NAME, SYSTEM_CATALOG_TABLE_NAME,
 };
 use common_catalog::format_full_table_name;
 use common_recordbatch::{RecordBatch, SendableRecordBatchStream};
@@ -27,7 +27,8 @@ use datatypes::prelude::ScalarVector;
 use datatypes::vectors::{BinaryVector, UInt8Vector};
 use futures_util::lock::Mutex;
 use snafu::{ensure, OptionExt, ResultExt};
-use table::engine::{EngineContext, TableEngineRef};
+use table::engine::manager::TableEngineManagerRef;
+use table::engine::EngineContext;
 use table::metadata::TableId;
 use table::requests::OpenTableRequest;
 use table::table::numbers::NumbersTable;
@@ -37,7 +38,8 @@ use table::TableRef;
 use crate::error::{
    self, CatalogNotFoundSnafu, IllegalManagerStateSnafu, OpenTableSnafu, ReadSystemCatalogSnafu,
    Result, SchemaExistsSnafu, SchemaNotFoundSnafu, SystemCatalogSnafu,
-    SystemCatalogTypeMismatchSnafu, TableExistsSnafu, TableNotFoundSnafu,
+    SystemCatalogTypeMismatchSnafu, TableEngineNotFoundSnafu, TableExistsSnafu, TableNotExistSnafu,
+    TableNotFoundSnafu,
 };
 use crate::local::memory::{MemoryCatalogManager, MemoryCatalogProvider, MemorySchemaProvider};
 use crate::system::{
@@ -55,7 +57,7 @@ use crate::{
 pub struct LocalCatalogManager {
    system: Arc<SystemCatalog>,
    catalogs: Arc<MemoryCatalogManager>,
-    engine: TableEngineRef,
+    engine_manager: TableEngineManagerRef,
    next_table_id: AtomicU32,
    init_lock: Mutex<bool>,
    register_lock: Mutex<()>,
@@ -63,19 +65,20 @@ pub struct LocalCatalogManager {
 }

 impl LocalCatalogManager {
-    /// Create a new [CatalogManager] with given user catalogs and table engine
-    pub async fn try_new(engine: TableEngineRef) -> Result<Self> {
+    /// Create a new [CatalogManager] with given user catalogs and mito engine
+    pub async fn try_new(engine_manager: TableEngineManagerRef) -> Result<Self> {
+        let engine = engine_manager
+            .engine(MITO_ENGINE)
+            .context(TableEngineNotFoundSnafu {
+                engine_name: MITO_ENGINE,
+            })?;
        let table = SystemCatalogTable::new(engine.clone()).await?;
        let memory_catalog_list = crate::local::memory::new_memory_catalog_list()?;
-        let system_catalog = Arc::new(SystemCatalog::new(
-            table,
-            memory_catalog_list.clone(),
-            engine.clone(),
-        ));
+        let system_catalog = Arc::new(SystemCatalog::new(table));
        Ok(Self {
            system: system_catalog,
            catalogs: memory_catalog_list,
-            engine,
+            engine_manager,
            next_table_id: AtomicU32::new(MIN_USER_TABLE_ID),
            init_lock: Mutex::new(false),
            register_lock: Mutex::new(()),
@@ -100,7 +103,14 @@ impl LocalCatalogManager {

        // Processing system table hooks
        let mut sys_table_requests = self.system_table_requests.lock().await;
-        handle_system_table_request(self, self.engine.clone(), &mut sys_table_requests).await?;
+        let engine = self
+            .engine_manager
+            .engine(MITO_ENGINE)
+            .context(TableEngineNotFoundSnafu {
+                engine_name: MITO_ENGINE,
+            })?;
+
+        handle_system_table_request(self, engine, &mut sys_table_requests).await?;
        Ok(())
    }

@@ -253,9 +263,14 @@ impl LocalCatalogManager {
            table_name: t.table_name.clone(),
            table_id: t.table_id,
        };
+        let engine = self
+            .engine_manager
+            .engine(&t.engine)
+            .context(TableEngineNotFoundSnafu {
+                engine_name: &t.engine,
+            })?;

-        let option = self
-            .engine
+        let option = engine
            .open_table(&context, request)
            .await
            .with_context(|_| OpenTableSnafu {
@@ -290,9 +305,7 @@ impl CatalogList for LocalCatalogManager {
    }

    fn catalog_names(&self) -> Result<Vec<String>> {
-        let mut res = self.catalogs.catalog_names()?;
-        res.push(SYSTEM_CATALOG_NAME.to_string());
-        Ok(res)
+        self.catalogs.catalog_names()
    }

    fn catalog(&self, name: &str) -> Result<Option<CatalogProviderRef>> {
@@ -345,7 +358,7 @@ impl CatalogManager for LocalCatalogManager {

        {
            let _lock = self.register_lock.lock().await;
-            if let Some(existing) = schema.table(&request.table_name)? {
+            if let Some(existing) = schema.table(&request.table_name).await? {
                if existing.table_info().ident.table_id != request.table_id {
                    error!(
                        "Unexpected table register request: {:?}, existing: {:?}",
@@ -364,6 +377,7 @@ impl CatalogManager for LocalCatalogManager {
                // Try to register table with same table id, just ignore.
                Ok(false)
            } else {
+                let engine = request.table.table_info().meta.engine.to_string();
                // table does not exist
                self.system
                    .register_table(
@@ -371,6 +385,7 @@ impl CatalogManager for LocalCatalogManager {
                        schema_name.clone(),
                        request.table_name.clone(),
                        request.table_id,
+                        engine,
                    )
                    .await?;
                schema.register_table(request.table_name, request.table)?;
@@ -404,6 +419,21 @@ impl CatalogManager for LocalCatalogManager {
                schema: schema_name,
            })?;

+        let _lock = self.register_lock.lock().await;
+        ensure!(
+            !schema.table_exist(&request.new_table_name)?,
+            TableExistsSnafu {
+                table: &request.new_table_name
+            }
+        );
+        let old_table = schema
+            .table(&request.table_name)
+            .await?
+            .context(TableNotExistSnafu {
+                table: &request.table_name,
+            })?;
+
+        let engine = old_table.table_info().meta.engine.to_string();
        // rename table in system catalog
        self.system
            .register_table(
@@ -411,11 +441,14 @@ impl CatalogManager for LocalCatalogManager {
                schema_name.clone(),
                request.new_table_name.clone(),
                request.table_id,
+                engine,
            )
            .await?;
-        Ok(schema
-            .rename_table(&request.table_name, request.new_table_name)
-            .is_ok())
+
+        let renamed = schema
+            .rename_table(&request.table_name, request.new_table_name.clone())
+            .is_ok();
+        Ok(renamed)
    }

    async fn deregister_table(&self, request: DeregisterTableRequest) -> Result<bool> {
@@ -434,9 +467,10 @@ impl CatalogManager for LocalCatalogManager {
            } = &request;
            let table_id = self
                .catalogs
-                .table(catalog, schema, table_name)?
+                .table(catalog, schema, table_name)
+                .await?
                .with_context(|| error::TableNotExistSnafu {
-                    table: format!("{catalog}.{schema}.{table_name}"),
+                    table: format_full_table_name(catalog, schema, table_name),
                })?
                .table_info()
                .ident
@@ -505,7 +539,7 @@ impl CatalogManager for LocalCatalogManager {
            .schema(schema)
    }

-    fn table(
+    async fn table(
        &self,
        catalog_name: &str,
        schema_name: &str,
@@ -521,7 +555,7 @@ impl CatalogManager for LocalCatalogManager {
                catalog: catalog_name,
                schema: schema_name,
            })?;
-        schema.table(table_name)
+        schema.table(table_name).await
    }
 }

@@ -529,6 +563,8 @@ impl CatalogManager for LocalCatalogManager {
 mod tests {
    use std::assert_matches::assert_matches;

+    use mito::engine::MITO_ENGINE;
+
    use super::*;
    use crate::system::{CatalogEntry, SchemaEntry};

@@ -540,6 +576,7 @@ mod tests {
                schema_name: "S1".to_string(),
                table_name: "T1".to_string(),
                table_id: 1,
+                engine: MITO_ENGINE.to_string(),
            }),
            Entry::Catalog(CatalogEntry {
                catalog_name: "C2".to_string(),
@@ -560,6 +597,7 @@ mod tests {
                schema_name: "S1".to_string(),
                table_name: "T2".to_string(),
                table_id: 2,
+                engine: MITO_ENGINE.to_string(),
            }),
        ];
        let res = LocalCatalogManager::sort_entries(vec);
--- a/src/catalog/src/local/memory.rs
+++ b/src/catalog/src/local/memory.rs
@@ -18,6 +18,7 @@ use std::collections::HashMap;
 use std::sync::atomic::{AtomicU32, Ordering};
 use std::sync::{Arc, RwLock};

+use async_trait::async_trait;
 use common_catalog::consts::MIN_USER_TABLE_ID;
 use common_telemetry::error;
 use snafu::{ensure, OptionExt};
@@ -155,16 +156,20 @@ impl CatalogManager for MemoryCatalogManager {
        }
    }

-    fn table(&self, catalog: &str, schema: &str, table_name: &str) -> Result<Option<TableRef>> {
-        let c = self.catalogs.read().unwrap();
-        let catalog = if let Some(c) = c.get(catalog) {
+    async fn table(
+        &self,
+        catalog: &str,
+        schema: &str,
+        table_name: &str,
+    ) -> Result<Option<TableRef>> {
+        let catalog = {
+            let c = self.catalogs.read().unwrap();
+            let Some(c) = c.get(catalog) else { return Ok(None) };
            c.clone()
-        } else {
-            return Ok(None);
        };
        match catalog.schema(schema)? {
            None => Ok(None),
-            Some(s) => s.table(table_name),
+            Some(s) => s.table(table_name).await,
        }
    }
 }
@@ -283,6 +288,7 @@ impl Default for MemorySchemaProvider {
    }
 }

+#[async_trait]
 impl SchemaProvider for MemorySchemaProvider {
    fn as_any(&self) -> &dyn Any {
        self
@@ -293,7 +299,7 @@ impl SchemaProvider for MemorySchemaProvider {
        Ok(tables.keys().cloned().collect())
    }

-    fn table(&self, name: &str) -> Result<Option<TableRef>> {
+    async fn table(&self, name: &str) -> Result<Option<TableRef>> {
        let tables = self.tables.read().unwrap();
        Ok(tables.get(name).cloned())
    }
@@ -318,16 +324,20 @@ impl SchemaProvider for MemorySchemaProvider {

    fn rename_table(&self, name: &str, new_name: String) -> Result<TableRef> {
        let mut tables = self.tables.write().unwrap();
-        if tables.get(name).is_some() {
-            let table = tables.remove(name).unwrap();
-            tables.insert(new_name, table.clone());
-            Ok(table)
-        } else {
-            TableNotFoundSnafu {
+        let Some(table) = tables.remove(name) else {
+            return TableNotFoundSnafu {
                table_info: name.to_string(),
            }
-            .fail()?
-        }
+                .fail()?;
+        };
+        let e = match tables.entry(new_name) {
+            Entry::Vacant(e) => e,
+            Entry::Occupied(e) => {
+                return TableExistsSnafu { table: e.key() }.fail();
+            }
+        };
+        e.insert(table.clone());
+        Ok(table)
    }

    fn deregister_table(&self, name: &str) -> Result<Option<TableRef>> {
@@ -355,8 +365,8 @@ mod tests {

    use super::*;

-    #[test]
-    fn test_new_memory_catalog_list() {
+    #[tokio::test]
+    async fn test_new_memory_catalog_list() {
        let catalog_list = new_memory_catalog_list().unwrap();
        let default_catalog = catalog_list.catalog(DEFAULT_CATALOG_NAME).unwrap().unwrap();

@@ -369,9 +379,9 @@ mod tests {
            .register_table("numbers".to_string(), Arc::new(NumbersTable::default()))
            .unwrap();

-        let table = default_schema.table("numbers").unwrap();
+        let table = default_schema.table("numbers").await.unwrap();
        assert!(table.is_some());
-        assert!(default_schema.table("not_exists").unwrap().is_none());
+        assert!(default_schema.table("not_exists").await.unwrap().is_none());
    }

    #[tokio::test]
@@ -390,7 +400,6 @@ mod tests {
        let other_table = NumbersTable::new(12);
        let result = provider.register_table(table_name.to_string(), Arc::new(other_table));
        let err = result.err().unwrap();
-        assert!(err.backtrace_opt().is_some());
        assert_eq!(StatusCode::TableAlreadyExists, err.status_code());
    }

@@ -419,7 +428,7 @@ mod tests {

        // test new table name exists
        assert!(provider.table_exist(new_table_name).unwrap());
-        let registered_table = provider.table(new_table_name).unwrap().unwrap();
+        let registered_table = provider.table(new_table_name).await.unwrap().unwrap();
        assert_eq!(
            registered_table.table_info().ident.table_id,
            test_table.table_info().ident.table_id
@@ -454,7 +463,7 @@ mod tests {
        assert!(schema.table_exist(table_name).unwrap());

        // rename table
-        let new_table_name = "numbers";
+        let new_table_name = "numbers_new";
        let rename_table_req = RenameTableRequest {
            catalog: DEFAULT_CATALOG_NAME.to_string(),
            schema: DEFAULT_SCHEMA_NAME.to_string(),
@@ -468,6 +477,7 @@ mod tests {

        let registered_table = catalog
            .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, new_table_name)
+            .await
            .unwrap()
            .unwrap();
        assert_eq!(registered_table.table_info().ident.table_id, table_id);
--- a/src/catalog/src/remote/manager.rs
+++ b/src/catalog/src/remote/manager.rs
@@ -13,31 +13,37 @@
 // limitations under the License.

 use std::any::Any;
-use std::collections::HashMap;
+use std::collections::{HashMap, HashSet};
 use std::pin::Pin;
 use std::sync::Arc;

 use arc_swap::ArcSwap;
 use async_stream::stream;
-use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, MIN_USER_TABLE_ID};
-use common_telemetry::{debug, info};
+use async_trait::async_trait;
+use common_catalog::consts::{
+    DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, MIN_USER_TABLE_ID, MITO_ENGINE,
+};
+use common_telemetry::{debug, error, info};
+use dashmap::DashMap;
 use futures::Stream;
 use futures_util::StreamExt;
+use parking_lot::RwLock;
 use snafu::{OptionExt, ResultExt};
-use table::engine::{EngineContext, TableEngineRef};
+use table::engine::manager::TableEngineManagerRef;
+use table::engine::EngineContext;
 use table::metadata::TableId;
 use table::requests::{CreateTableRequest, OpenTableRequest};
-use table::table::numbers::NumbersTable;
 use table::TableRef;
 use tokio::sync::Mutex;

 use crate::error::{
    CatalogNotFoundSnafu, CreateTableSnafu, InvalidCatalogValueSnafu, OpenTableSnafu, Result,
-    SchemaNotFoundSnafu, TableExistsSnafu, UnimplementedSnafu,
+    SchemaNotFoundSnafu, TableEngineNotFoundSnafu, TableExistsSnafu, UnimplementedSnafu,
 };
 use crate::helper::{
    build_catalog_prefix, build_schema_prefix, build_table_global_prefix, CatalogKey, CatalogValue,
    SchemaKey, SchemaValue, TableGlobalKey, TableGlobalValue, TableRegionalKey, TableRegionalValue,
+    CATALOG_KEY_PREFIX,
 };
 use crate::remote::{Kv, KvBackendRef};
 use crate::{
@@ -50,21 +56,19 @@ use crate::{
 pub struct RemoteCatalogManager {
    node_id: u64,
    backend: KvBackendRef,
-    catalogs: Arc<ArcSwap<HashMap<String, CatalogProviderRef>>>,
-    engine: TableEngineRef,
+    catalogs: Arc<RwLock<DashMap<String, CatalogProviderRef>>>,
+    engine_manager: TableEngineManagerRef,
    system_table_requests: Mutex<Vec<RegisterSystemTableRequest>>,
-    mutex: Arc<Mutex<()>>,
 }

 impl RemoteCatalogManager {
-    pub fn new(engine: TableEngineRef, node_id: u64, backend: KvBackendRef) -> Self {
+    pub fn new(engine_manager: TableEngineManagerRef, node_id: u64, backend: KvBackendRef) -> Self {
        Self {
-            engine,
+            engine_manager,
            node_id,
            backend,
            catalogs: Default::default(),
            system_table_requests: Default::default(),
-            mutex: Default::default(),
        }
    }

@@ -108,9 +112,13 @@ impl RemoteCatalogManager {
                    debug!("Ignoring non-catalog key: {}", String::from_utf8_lossy(&k));
                    continue;
                }
-                let key = CatalogKey::parse(&String::from_utf8_lossy(&k))
-                    .context(InvalidCatalogValueSnafu)?;
-                yield Ok(key)
+
+                let catalog_key = String::from_utf8_lossy(&k);
+                if let Ok(key) = CatalogKey::parse(&catalog_key) {
+                    yield Ok(key)
+                } else {
+                    error!("Invalid catalog key: {:?}", catalog_key);
+                }
            }
        }))
    }
@@ -180,7 +188,9 @@ impl RemoteCatalogManager {
        let max_table_id = MIN_USER_TABLE_ID - 1;

        // initiate default catalog and schema
-        let default_catalog = self.initiate_default_catalog().await?;
+        let default_catalog = self
+            .create_catalog_and_schema(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME)
+            .await?;
        res.insert(DEFAULT_CATALOG_NAME.to_string(), default_catalog);
        info!("Default catalog and schema registered");

@@ -260,13 +270,19 @@ impl RemoteCatalogManager {
        Ok(())
    }

-    async fn initiate_default_catalog(&self) -> Result<CatalogProviderRef> {
-        let default_catalog = self.new_catalog_provider(DEFAULT_CATALOG_NAME);
-        let default_schema = self.new_schema_provider(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME);
-        default_catalog.register_schema(DEFAULT_SCHEMA_NAME.to_string(), default_schema.clone())?;
+    pub async fn create_catalog_and_schema(
+        &self,
+        catalog_name: &str,
+        schema_name: &str,
+    ) -> Result<CatalogProviderRef> {
+        let schema_provider = self.new_schema_provider(catalog_name, schema_name);
+
+        let catalog_provider = self.new_catalog_provider(catalog_name);
+        catalog_provider.register_schema(schema_name.to_string(), schema_provider.clone())?;
+
        let schema_key = SchemaKey {
-            schema_name: DEFAULT_SCHEMA_NAME.to_string(),
-            catalog_name: DEFAULT_CATALOG_NAME.to_string(),
+            catalog_name: catalog_name.to_string(),
+            schema_name: schema_name.to_string(),
        }
        .to_string();
        self.backend
@@ -277,10 +293,10 @@ impl RemoteCatalogManager {
                    .context(InvalidCatalogValueSnafu)?,
            )
            .await?;
-        info!("Registered default schema");
+        info!("Created schema '{schema_key}'");

        let catalog_key = CatalogKey {
-            catalog_name: DEFAULT_CATALOG_NAME.to_string(),
+            catalog_name: catalog_name.to_string(),
        }
        .to_string();
        self.backend
@@ -291,8 +307,8 @@ impl RemoteCatalogManager {
                    .context(InvalidCatalogValueSnafu)?,
            )
            .await?;
-        info!("Registered default catalog");
-        Ok(default_catalog)
+        info!("Created catalog '{catalog_key}");
+        Ok(catalog_provider)
    }

    async fn open_or_create_table(
@@ -325,8 +341,13 @@ impl RemoteCatalogManager {
            table_name: table_name.clone(),
            table_id,
        };
-        match self
-            .engine
+        let engine = self
+            .engine_manager
+            .engine(&table_info.meta.engine)
+            .context(TableEngineNotFoundSnafu {
+                engine_name: &table_info.meta.engine,
+            })?;
+        match engine
            .open_table(&context, request)
            .await
            .with_context(|_| OpenTableSnafu {
@@ -357,9 +378,10 @@ impl RemoteCatalogManager {
                    primary_key_indices: meta.primary_key_indices.clone(),
                    create_if_not_exists: true,
                    table_options: meta.options.clone(),
+                    engine: engine.name().to_string(),
                };

-                self.engine
+                engine
                    .create_table(&context, req)
                    .await
                    .context(CreateTableSnafu {
@@ -381,21 +403,25 @@ impl CatalogManager for RemoteCatalogManager {
            "Initialized catalogs: {:?}",
            catalogs.keys().cloned().collect::<Vec<_>>()
        );
-        self.catalogs.store(Arc::new(catalogs));
+
+        {
+            let self_catalogs = self.catalogs.read();
+            catalogs.into_iter().for_each(|(k, v)| {
+                self_catalogs.insert(k, v);
+            });
+        }
+
        info!("Max table id allocated: {}", max_table_id);

        let mut system_table_requests = self.system_table_requests.lock().await;
-        handle_system_table_request(self, self.engine.clone(), &mut system_table_requests).await?;
+        let engine = self
+            .engine_manager
+            .engine(MITO_ENGINE)
+            .context(TableEngineNotFoundSnafu {
+                engine_name: MITO_ENGINE,
+            })?;
+        handle_system_table_request(self, engine, &mut system_table_requests).await?;
        info!("All system table opened");
-
-        self.catalog(DEFAULT_CATALOG_NAME)
-            .unwrap()
-            .unwrap()
-            .schema(DEFAULT_SCHEMA_NAME)
-            .unwrap()
-            .unwrap()
-            .register_table("numbers".to_string(), Arc::new(NumbersTable::default()))
-            .unwrap();
        Ok(())
    }

@@ -468,7 +494,7 @@ impl CatalogManager for RemoteCatalogManager {
            .schema(schema)
    }

-    fn table(
+    async fn table(
        &self,
        catalog_name: &str,
        schema_name: &str,
@@ -483,7 +509,7 @@ impl CatalogManager for RemoteCatalogManager {
                catalog: catalog_name,
                schema: schema_name,
            })?;
-        schema.table(table_name)
+        schema.table(table_name).await
    }
 }

@@ -499,12 +525,10 @@ impl CatalogList for RemoteCatalogManager {
    ) -> Result<Option<CatalogProviderRef>> {
        let key = self.build_catalog_key(&name).to_string();
        let backend = self.backend.clone();
-        let mutex = self.mutex.clone();
        let catalogs = self.catalogs.clone();

        std::thread::spawn(|| {
            common_runtime::block_on_write(async move {
-                let _guard = mutex.lock().await;
                backend
                    .set(
                        key.as_bytes(),
@@ -513,11 +537,10 @@ impl CatalogList for RemoteCatalogManager {
                            .context(InvalidCatalogValueSnafu)?,
                    )
                    .await?;
-                let prev_catalogs = catalogs.load();
-                let mut new_catalogs = HashMap::with_capacity(prev_catalogs.len() + 1);
-                new_catalogs.clone_from(&prev_catalogs);
-                let prev = new_catalogs.insert(name, catalog);
-                catalogs.store(Arc::new(new_catalogs));
+
+                let catalogs = catalogs.read();
+                let prev = catalogs.insert(name, catalog.clone());
+
                Ok(prev)
            })
        })
@@ -527,12 +550,65 @@ impl CatalogList for RemoteCatalogManager {

    /// List all catalogs from metasrv
    fn catalog_names(&self) -> Result<Vec<String>> {
-        Ok(self.catalogs.load().keys().cloned().collect::<Vec<_>>())
+        let catalogs = self.catalogs.read();
+        Ok(catalogs.iter().map(|k| k.key().to_string()).collect())
    }

    /// Read catalog info of given name from metasrv.
    fn catalog(&self, name: &str) -> Result<Option<CatalogProviderRef>> {
-        Ok(self.catalogs.load().get(name).cloned())
+        {
+            let catalogs = self.catalogs.read();
+            let catalog = catalogs.get(name);
+
+            if let Some(catalog) = catalog {
+                return Ok(Some(catalog.clone()));
+            }
+        }
+
+        let catalogs = self.catalogs.write();
+
+        let catalog = catalogs.get(name);
+        if let Some(catalog) = catalog {
+            return Ok(Some(catalog.clone()));
+        }
+
+        // It's for lack of incremental catalog syncing between datanode and meta. Here we fetch catalog
+        // from meta on demand. This can be removed when incremental catalog syncing is done in datanode.
+
+        let backend = self.backend.clone();
+
+        let catalogs_from_meta: HashSet<String> = std::thread::spawn(|| {
+            common_runtime::block_on_read(async move {
+                let mut stream = backend.range(CATALOG_KEY_PREFIX.as_bytes());
+                let mut catalogs = HashSet::new();
+
+                while let Some(catalog) = stream.next().await {
+                    if let Ok(catalog) = catalog {
+                        let catalog_key = String::from_utf8_lossy(&catalog.0);
+
+                        if let Ok(key) = CatalogKey::parse(&catalog_key) {
+                            catalogs.insert(key.catalog_name);
+                        }
+                    }
+                }
+
+                catalogs
+            })
+        })
+        .join()
+        .unwrap();
+
+        catalogs.retain(|catalog_name, _| catalogs_from_meta.get(catalog_name).is_some());
+
+        for catalog in catalogs_from_meta {
+            catalogs
+                .entry(catalog.clone())
+                .or_insert(self.new_catalog_provider(&catalog));
+        }
+
+        let catalog = catalogs.get(name);
+
+        Ok(catalog.as_deref().cloned())
    }
 }

@@ -692,6 +768,7 @@ impl RemoteSchemaProvider {
    }
 }

+#[async_trait]
 impl SchemaProvider for RemoteSchemaProvider {
    fn as_any(&self) -> &dyn Any {
        self
@@ -701,7 +778,7 @@ impl SchemaProvider for RemoteSchemaProvider {
        Ok(self.tables.load().keys().cloned().collect::<Vec<_>>())
    }

-    fn table(&self, name: &str) -> Result<Option<TableRef>> {
+    async fn table(&self, name: &str) -> Result<Option<TableRef>> {
        Ok(self.tables.load().get(name).cloned())
    }

--- a/src/catalog/src/schema.rs
+++ b/src/catalog/src/schema.rs
@@ -15,11 +15,13 @@
 use std::any::Any;
 use std::sync::Arc;

+use async_trait::async_trait;
 use table::TableRef;

-use crate::error::Result;
+use crate::error::{NotSupportedSnafu, Result};

 /// Represents a schema, comprising a number of named tables.
+#[async_trait]
 pub trait SchemaProvider: Sync + Send {
    /// Returns the schema provider as [`Any`](std::any::Any)
    /// so that it can be downcast to a specific implementation.
@@ -29,19 +31,34 @@ pub trait SchemaProvider: Sync + Send {
    fn table_names(&self) -> Result<Vec<String>>;

    /// Retrieves a specific table from the schema by name, provided it exists.
-    fn table(&self, name: &str) -> Result<Option<TableRef>>;
+    async fn table(&self, name: &str) -> Result<Option<TableRef>>;

    /// If supported by the implementation, adds a new table to this schema.
    /// If a table of the same name existed before, it returns "Table already exists" error.
-    fn register_table(&self, name: String, table: TableRef) -> Result<Option<TableRef>>;
+    fn register_table(&self, name: String, _table: TableRef) -> Result<Option<TableRef>> {
+        NotSupportedSnafu {
+            op: format!("register_table({name}, <table>)"),
+        }
+        .fail()
+    }

    /// If supported by the implementation, renames an existing table from this schema and returns it.
    /// If no table of that name exists, returns "Table not found" error.
-    fn rename_table(&self, name: &str, new_name: String) -> Result<TableRef>;
+    fn rename_table(&self, name: &str, new_name: String) -> Result<TableRef> {
+        NotSupportedSnafu {
+            op: format!("rename_table({name}, {new_name})"),
+        }
+        .fail()
+    }

    /// If supported by the implementation, removes an existing table from this schema and returns it.
    /// If no table of that name exists, returns Ok(None).
-    fn deregister_table(&self, name: &str) -> Result<Option<TableRef>>;
+    fn deregister_table(&self, name: &str) -> Result<Option<TableRef>> {
+        NotSupportedSnafu {
+            op: format!("deregister_table({name})"),
+        }
+        .fail()
+    }

    /// If supported by the implementation, checks the table exist in the schema provider or not.
    /// If no matched table in the schema provider, return false.
--- a/src/catalog/src/system.rs
+++ b/src/catalog/src/system.rs
@@ -17,8 +17,8 @@ use std::collections::HashMap;
 use std::sync::Arc;

 use common_catalog::consts::{
-    DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, SYSTEM_CATALOG_NAME,
-    SYSTEM_CATALOG_TABLE_ID, SYSTEM_CATALOG_TABLE_NAME,
+    DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, INFORMATION_SCHEMA_NAME, MITO_ENGINE,
+    SYSTEM_CATALOG_NAME, SYSTEM_CATALOG_TABLE_ID, SYSTEM_CATALOG_TABLE_NAME,
 };
 use common_query::logical_plan::Expr;
 use common_query::physical_plan::{PhysicalPlanRef, SessionContext};
@@ -32,7 +32,9 @@ use serde::{Deserialize, Serialize};
 use snafu::{ensure, OptionExt, ResultExt};
 use table::engine::{EngineContext, TableEngineRef};
 use table::metadata::{TableId, TableInfoRef};
-use table::requests::{CreateTableRequest, DeleteRequest, InsertRequest, OpenTableRequest};
+use table::requests::{
+    CreateTableRequest, DeleteRequest, InsertRequest, OpenTableRequest, TableOptions,
+};
 use table::{Table, TableRef};

 use crate::error::{
@@ -109,7 +111,8 @@ impl SystemCatalogTable {
                region_numbers: vec![0],
                primary_key_indices: vec![ENTRY_TYPE_INDEX, KEY_INDEX],
                create_if_not_exists: true,
-                table_options: HashMap::new(),
+                table_options: TableOptions::default(),
+                engine: engine.name().to_string(),
            };

            let table = engine
@@ -192,12 +195,13 @@ pub fn build_table_insert_request(
    schema: String,
    table_name: String,
    table_id: TableId,
+    engine: String,
 ) -> InsertRequest {
    let entry_key = format_table_entry_key(&catalog, &schema, table_id);
    build_insert_request(
        EntryType::Table,
        entry_key.as_bytes(),
-        serde_json::to_string(&TableEntryValue { table_name })
+        serde_json::to_string(&TableEntryValue { table_name, engine })
            .unwrap()
            .as_bytes(),
    )
@@ -217,7 +221,7 @@ fn build_primary_key_columns(entry_type: EntryType, key: &[u8]) -> HashMap<Strin
    let mut m = HashMap::with_capacity(3);
    m.insert(
        "entry_type".to_string(),
-        Arc::new(UInt8Vector::from_slice(&[entry_type as u8])) as _,
+        Arc::new(UInt8Vector::from_slice([entry_type as u8])) as _,
    );
    m.insert(
        "key".to_string(),
@@ -226,7 +230,7 @@ fn build_primary_key_columns(entry_type: EntryType, key: &[u8]) -> HashMap<Strin
    // Timestamp in key part is intentionally left to 0
    m.insert(
        "timestamp".to_string(),
-        Arc::new(TimestampMillisecondVector::from_slice(&[0])) as _,
+        Arc::new(TimestampMillisecondVector::from_slice([0])) as _,
    );
    m
 }
@@ -256,12 +260,12 @@ pub fn build_insert_request(entry_type: EntryType, key: &[u8], value: &[u8]) ->
    let now = util::current_time_millis();
    columns_values.insert(
        "gmt_created".to_string(),
-        Arc::new(TimestampMillisecondVector::from_slice(&[now])) as _,
+        Arc::new(TimestampMillisecondVector::from_slice([now])) as _,
    );

    columns_values.insert(
        "gmt_modified".to_string(),
-        Arc::new(TimestampMillisecondVector::from_slice(&[now])) as _,
+        Arc::new(TimestampMillisecondVector::from_slice([now])) as _,
    );

    InsertRequest {
@@ -328,6 +332,7 @@ pub fn decode_system_catalog(
                schema_name: table_parts[1].to_string(),
                table_name: table_meta.table_name,
                table_id,
+                engine: table_meta.engine,
            }))
        }
    }
@@ -383,27 +388,35 @@ pub struct TableEntry {
    pub schema_name: String,
    pub table_name: String,
    pub table_id: TableId,
+    pub engine: String,
 }

 #[derive(Debug, Serialize, Deserialize, PartialEq, Eq)]
 pub struct TableEntryValue {
    pub table_name: String,
+
+    #[serde(default = "mito_engine")]
+    pub engine: String,
+}
+
+fn mito_engine() -> String {
+    MITO_ENGINE.to_string()
 }

 #[cfg(test)]
 mod tests {
    use common_recordbatch::RecordBatches;
+    use common_test_util::temp_dir::{create_temp_dir, TempDir};
    use datatypes::value::Value;
    use log_store::NoopLogStore;
    use mito::config::EngineConfig;
-    use mito::engine::MitoEngine;
+    use mito::engine::{MitoEngine, MITO_ENGINE};
    use object_store::ObjectStore;
    use storage::compaction::noop::NoopCompactionScheduler;
    use storage::config::EngineConfig as StorageEngineConfig;
    use storage::EngineImpl;
    use table::metadata::TableType;
    use table::metadata::TableType::Base;
-    use tempdir::TempDir;

    use super::*;

@@ -478,13 +491,11 @@ mod tests {
    }

    pub async fn prepare_table_engine() -> (TempDir, TableEngineRef) {
-        let dir = TempDir::new("system-table-test").unwrap();
+        let dir = create_temp_dir("system-table-test");
        let store_dir = dir.path().to_string_lossy();
-        let accessor = object_store::backend::fs::Builder::default()
-            .root(&store_dir)
-            .build()
-            .unwrap();
-        let object_store = ObjectStore::new(accessor);
+        let mut builder = object_store::services::Fs::default();
+        builder.root(&store_dir);
+        let object_store = ObjectStore::new(builder).unwrap().finish();
        let noop_compaction_scheduler = Arc::new(NoopCompactionScheduler::default());
        let table_engine = Arc::new(MitoEngine::new(
            EngineConfig::default(),
@@ -528,6 +539,7 @@ mod tests {
            DEFAULT_SCHEMA_NAME.to_string(),
            "my_table".to_string(),
            1,
+            MITO_ENGINE.to_string(),
        );
        let result = catalog_table.insert(table_insertion).await.unwrap();
        assert_eq!(result, 1);
@@ -548,6 +560,7 @@ mod tests {
            schema_name: DEFAULT_SCHEMA_NAME.to_string(),
            table_name: "my_table".to_string(),
            table_id: 1,
+            engine: MITO_ENGINE.to_string(),
        });
        assert_eq!(entry, expected);

--- a/src/catalog/src/table_source.rs
+++ b/src/catalog/src/table_source.rs
@@ -0,0 +1,190 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::collections::HashMap;
+use std::sync::Arc;
+
+use common_catalog::consts::INFORMATION_SCHEMA_NAME;
+use common_catalog::format_full_table_name;
+use datafusion::common::{ResolvedTableReference, TableReference};
+use datafusion::datasource::provider_as_source;
+use datafusion::logical_expr::TableSource;
+use session::context::QueryContext;
+use snafu::{ensure, OptionExt};
+use table::table::adapter::DfTableProviderAdapter;
+
+use crate::error::{
+    CatalogNotFoundSnafu, QueryAccessDeniedSnafu, Result, SchemaNotFoundSnafu, TableNotExistSnafu,
+};
+use crate::information_schema::InformationSchemaProvider;
+use crate::CatalogListRef;
+
+pub struct DfTableSourceProvider {
+    catalog_list: CatalogListRef,
+    resolved_tables: HashMap<String, Arc<dyn TableSource>>,
+    disallow_cross_schema_query: bool,
+    default_catalog: String,
+    default_schema: String,
+}
+
+impl DfTableSourceProvider {
+    pub fn new(
+        catalog_list: CatalogListRef,
+        disallow_cross_schema_query: bool,
+        query_ctx: &QueryContext,
+    ) -> Self {
+        Self {
+            catalog_list,
+            disallow_cross_schema_query,
+            resolved_tables: HashMap::new(),
+            default_catalog: query_ctx.current_catalog(),
+            default_schema: query_ctx.current_schema(),
+        }
+    }
+
+    pub fn resolve_table_ref<'a>(
+        &'a self,
+        table_ref: TableReference<'a>,
+    ) -> Result<ResolvedTableReference<'a>> {
+        if self.disallow_cross_schema_query {
+            match &table_ref {
+                TableReference::Bare { .. } => (),
+                TableReference::Partial { schema, .. } => {
+                    ensure!(
+                        schema.as_ref() == self.default_schema,
+                        QueryAccessDeniedSnafu {
+                            catalog: &self.default_catalog,
+                            schema: schema.as_ref(),
+                        }
+                    );
+                }
+                TableReference::Full {
+                    catalog, schema, ..
+                } => {
+                    ensure!(
+                        catalog.as_ref() == self.default_catalog
+                            && schema.as_ref() == self.default_schema,
+                        QueryAccessDeniedSnafu {
+                            catalog: catalog.as_ref(),
+                            schema: schema.as_ref()
+                        }
+                    );
+                }
+            };
+        }
+
+        Ok(table_ref.resolve(&self.default_catalog, &self.default_schema))
+    }
+
+    pub async fn resolve_table(
+        &mut self,
+        table_ref: TableReference<'_>,
+    ) -> Result<Arc<dyn TableSource>> {
+        let table_ref = self.resolve_table_ref(table_ref)?;
+
+        let resolved_name = table_ref.to_string();
+        if let Some(table) = self.resolved_tables.get(&resolved_name) {
+            return Ok(table.clone());
+        }
+
+        let catalog_name = table_ref.catalog.as_ref();
+        let schema_name = table_ref.schema.as_ref();
+        let table_name = table_ref.table.as_ref();
+
+        let schema = if schema_name != INFORMATION_SCHEMA_NAME {
+            let catalog = self
+                .catalog_list
+                .catalog(catalog_name)?
+                .context(CatalogNotFoundSnafu { catalog_name })?;
+            catalog.schema(schema_name)?.context(SchemaNotFoundSnafu {
+                catalog: catalog_name,
+                schema: schema_name,
+            })?
+        } else {
+            let catalog_provider = self
+                .catalog_list
+                .catalog(catalog_name)?
+                .context(CatalogNotFoundSnafu { catalog_name })?;
+            Arc::new(InformationSchemaProvider::new(
+                catalog_name.to_string(),
+                catalog_provider,
+            ))
+        };
+        let table = schema
+            .table(table_name)
+            .await?
+            .with_context(|| TableNotExistSnafu {
+                table: format_full_table_name(catalog_name, schema_name, table_name),
+            })?;
+
+        let table = DfTableProviderAdapter::new(table);
+        let table = provider_as_source(Arc::new(table));
+        self.resolved_tables.insert(resolved_name, table.clone());
+        Ok(table)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::borrow::Cow;
+
+    use session::context::QueryContext;
+
+    use super::*;
+    use crate::local::MemoryCatalogManager;
+
+    #[test]
+    fn test_validate_table_ref() {
+        let query_ctx = &QueryContext::with("greptime", "public");
+
+        let table_provider =
+            DfTableSourceProvider::new(Arc::new(MemoryCatalogManager::default()), true, query_ctx);
+
+        let table_ref = TableReference::Bare {
+            table: Cow::Borrowed("table_name"),
+        };
+        let result = table_provider.resolve_table_ref(table_ref);
+        assert!(result.is_ok());
+
+        let table_ref = TableReference::Partial {
+            schema: Cow::Borrowed("public"),
+            table: Cow::Borrowed("table_name"),
+        };
+        let result = table_provider.resolve_table_ref(table_ref);
+        assert!(result.is_ok());
+
+        let table_ref = TableReference::Partial {
+            schema: Cow::Borrowed("wrong_schema"),
+            table: Cow::Borrowed("table_name"),
+        };
+        let result = table_provider.resolve_table_ref(table_ref);
+        assert!(result.is_err());
+
+        let table_ref = TableReference::Full {
+            catalog: Cow::Borrowed("greptime"),
+            schema: Cow::Borrowed("public"),
+            table: Cow::Borrowed("table_name"),
+        };
+        let result = table_provider.resolve_table_ref(table_ref);
+        assert!(result.is_ok());
+
+        let table_ref = TableReference::Full {
+            catalog: Cow::Borrowed("wrong_catalog"),
+            schema: Cow::Borrowed("public"),
+            table: Cow::Borrowed("table_name"),
+        };
+        let result = table_provider.resolve_table_ref(table_ref);
+        assert!(result.is_err());
+    }
+}
--- a/src/catalog/src/tables.rs
+++ b/src/catalog/src/tables.rs
@@ -15,27 +15,12 @@
 // The `tables` table in system catalog keeps a record of all tables created by user.

 use std::any::Any;
-use std::pin::Pin;
 use std::sync::Arc;
-use std::task::{Context, Poll};

-use async_stream::stream;
+use async_trait::async_trait;
 use common_catalog::consts::{INFORMATION_SCHEMA_NAME, SYSTEM_CATALOG_TABLE_NAME};
-use common_error::ext::BoxedError;
-use common_query::logical_plan::Expr;
-use common_query::physical_plan::PhysicalPlanRef;
-use common_recordbatch::error::Result as RecordBatchResult;
-use common_recordbatch::{RecordBatch, RecordBatchStream};
-use datatypes::prelude::{ConcreteDataType, DataType};
-use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
-use datatypes::value::ValueRef;
-use datatypes::vectors::VectorRef;
-use futures::Stream;
 use snafu::ResultExt;
-use table::engine::TableEngineRef;
-use table::error::TablesRecordBatchSnafu;
-use table::metadata::{TableId, TableInfoRef};
-use table::table::scan::SimpleTableScan;
+use table::metadata::TableId;
 use table::{Table, TableRef};

 use crate::error::{self, Error, InsertCatalogRecordSnafu, Result as CatalogResult};
@@ -43,204 +28,32 @@ use crate::system::{
    build_schema_insert_request, build_table_deletion_request, build_table_insert_request,
    SystemCatalogTable,
 };
-use crate::{
-    CatalogListRef, CatalogProvider, DeregisterTableRequest, SchemaProvider, SchemaProviderRef,
-};
-
-/// Tables holds all tables created by user.
-pub struct Tables {
-    schema: SchemaRef,
-    catalogs: CatalogListRef,
-    engine_name: String,
-}
-
-impl Tables {
-    pub fn new(catalogs: CatalogListRef, engine_name: String) -> Self {
-        Self {
-            schema: Arc::new(build_schema_for_tables()),
-            catalogs,
-            engine_name,
-        }
-    }
-}
-
-#[async_trait::async_trait]
-impl Table for Tables {
-    fn as_any(&self) -> &dyn Any {
-        self
-    }
-
-    fn schema(&self) -> SchemaRef {
-        self.schema.clone()
-    }
-
-    fn table_info(&self) -> TableInfoRef {
-        unreachable!("Tables does not support table_info method")
-    }
-
-    async fn scan(
-        &self,
-        _projection: Option<&Vec<usize>>,
-        _filters: &[Expr],
-        _limit: Option<usize>,
-    ) -> table::error::Result<PhysicalPlanRef> {
-        let catalogs = self.catalogs.clone();
-        let schema_ref = self.schema.clone();
-        let engine_name = self.engine_name.clone();
-
-        let stream = stream!({
-            for catalog_name in catalogs
-                .catalog_names()
-                .map_err(BoxedError::new)
-                .context(TablesRecordBatchSnafu)?
-            {
-                let catalog = catalogs
-                    .catalog(&catalog_name)
-                    .map_err(BoxedError::new)
-                    .context(TablesRecordBatchSnafu)?
-                    .unwrap();
-                for schema_name in catalog
-                    .schema_names()
-                    .map_err(BoxedError::new)
-                    .context(TablesRecordBatchSnafu)?
-                {
-                    let mut tables_in_schema = Vec::with_capacity(
-                        catalog
-                            .schema_names()
-                            .map_err(BoxedError::new)
-                            .context(TablesRecordBatchSnafu)?
-                            .len(),
-                    );
-                    let schema = catalog
-                        .schema(&schema_name)
-                        .map_err(BoxedError::new)
-                        .context(TablesRecordBatchSnafu)?
-                        .unwrap();
-                    for table_name in schema
-                        .table_names()
-                        .map_err(BoxedError::new)
-                        .context(TablesRecordBatchSnafu)?
-                    {
-                        tables_in_schema.push(table_name);
-                    }
-
-                    let vec = tables_to_record_batch(
-                        &catalog_name,
-                        &schema_name,
-                        tables_in_schema,
-                        &engine_name,
-                    );
-                    let record_batch_res = RecordBatch::new(schema_ref.clone(), vec);
-                    yield record_batch_res;
-                }
-            }
-        });
-
-        let stream = Box::pin(TablesRecordBatchStream {
-            schema: self.schema.clone(),
-            stream: Box::pin(stream),
-        });
-        Ok(Arc::new(SimpleTableScan::new(stream)))
-    }
-}
-
-/// Convert tables info to `RecordBatch`.
-fn tables_to_record_batch(
-    catalog_name: &str,
-    schema_name: &str,
-    table_names: Vec<String>,
-    engine: &str,
-) -> Vec<VectorRef> {
-    let mut catalog_vec =
-        ConcreteDataType::string_datatype().create_mutable_vector(table_names.len());
-    let mut schema_vec =
-        ConcreteDataType::string_datatype().create_mutable_vector(table_names.len());
-    let mut table_name_vec =
-        ConcreteDataType::string_datatype().create_mutable_vector(table_names.len());
-    let mut engine_vec =
-        ConcreteDataType::string_datatype().create_mutable_vector(table_names.len());
-
-    for table_name in table_names {
-        // Safety: All these vectors are string type.
-        catalog_vec.push_value_ref(ValueRef::String(catalog_name));
-        schema_vec.push_value_ref(ValueRef::String(schema_name));
-        table_name_vec.push_value_ref(ValueRef::String(&table_name));
-        engine_vec.push_value_ref(ValueRef::String(engine));
-    }
-
-    vec![
-        catalog_vec.to_vector(),
-        schema_vec.to_vector(),
-        table_name_vec.to_vector(),
-        engine_vec.to_vector(),
-    ]
-}
-
-pub struct TablesRecordBatchStream {
-    schema: SchemaRef,
-    stream: Pin<Box<dyn Stream<Item = RecordBatchResult<RecordBatch>> + Send>>,
-}
-
-impl Stream for TablesRecordBatchStream {
-    type Item = RecordBatchResult<RecordBatch>;
-
-    fn poll_next(mut self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
-        Pin::new(&mut self.stream).poll_next(cx)
-    }
-}
-
-impl RecordBatchStream for TablesRecordBatchStream {
-    fn schema(&self) -> SchemaRef {
-        self.schema.clone()
-    }
-}
+use crate::{CatalogProvider, DeregisterTableRequest, SchemaProvider, SchemaProviderRef};

 pub struct InformationSchema {
-    pub tables: Arc<Tables>,
    pub system: Arc<SystemCatalogTable>,
 }

+#[async_trait]
 impl SchemaProvider for InformationSchema {
    fn as_any(&self) -> &dyn Any {
        self
    }

    fn table_names(&self) -> Result<Vec<String>, Error> {
-        Ok(vec![
-            "tables".to_string(),
-            SYSTEM_CATALOG_TABLE_NAME.to_string(),
-        ])
+        Ok(vec![SYSTEM_CATALOG_TABLE_NAME.to_string()])
    }

-    fn table(&self, name: &str) -> Result<Option<TableRef>, Error> {
-        if name.eq_ignore_ascii_case("tables") {
-            Ok(Some(self.tables.clone()))
-        } else if name.eq_ignore_ascii_case(SYSTEM_CATALOG_TABLE_NAME) {
+    async fn table(&self, name: &str) -> Result<Option<TableRef>, Error> {
+        if name.eq_ignore_ascii_case(SYSTEM_CATALOG_TABLE_NAME) {
            Ok(Some(self.system.clone()))
        } else {
            Ok(None)
        }
    }

-    fn register_table(
-        &self,
-        _name: String,
-        _table: TableRef,
-    ) -> crate::error::Result<Option<TableRef>> {
-        panic!("System catalog & schema does not support register table")
-    }
-
-    fn rename_table(&self, _name: &str, _new_name: String) -> crate::error::Result<TableRef> {
-        unimplemented!("System catalog & schema does not support rename table")
-    }
-
-    fn deregister_table(&self, _name: &str) -> crate::error::Result<Option<TableRef>> {
-        panic!("System catalog & schema does not support deregister table")
-    }
-
    fn table_exist(&self, name: &str) -> Result<bool, Error> {
-        Ok(name.eq_ignore_ascii_case("tables")
-            || name.eq_ignore_ascii_case(SYSTEM_CATALOG_TABLE_NAME))
+        Ok(name.eq_ignore_ascii_case(SYSTEM_CATALOG_TABLE_NAME))
    }
 }

@@ -249,13 +62,8 @@ pub struct SystemCatalog {
 }

 impl SystemCatalog {
-    pub fn new(
-        system: SystemCatalogTable,
-        catalogs: CatalogListRef,
-        engine: TableEngineRef,
-    ) -> Self {
+    pub(crate) fn new(system: SystemCatalogTable) -> Self {
        let schema = InformationSchema {
-            tables: Arc::new(Tables::new(catalogs, engine.name().to_string())),
            system: Arc::new(system),
        };
        Self {
@@ -269,8 +77,9 @@ impl SystemCatalog {
        schema: String,
        table_name: String,
        table_id: TableId,
+        engine: String,
    ) -> crate::error::Result<usize> {
-        let request = build_table_insert_request(catalog, schema, table_name, table_id);
+        let request = build_table_insert_request(catalog, schema, table_name, table_id, engine);
        self.information_schema
            .system
            .insert(request)
@@ -332,104 +141,3 @@ impl CatalogProvider for SystemCatalog {
        }
    }
 }
-
-fn build_schema_for_tables() -> Schema {
-    let cols = vec![
-        ColumnSchema::new(
-            "catalog".to_string(),
-            ConcreteDataType::string_datatype(),
-            false,
-        ),
-        ColumnSchema::new(
-            "schema".to_string(),
-            ConcreteDataType::string_datatype(),
-            false,
-        ),
-        ColumnSchema::new(
-            "table_name".to_string(),
-            ConcreteDataType::string_datatype(),
-            false,
-        ),
-        ColumnSchema::new(
-            "engine".to_string(),
-            ConcreteDataType::string_datatype(),
-            false,
-        ),
-    ];
-    Schema::new(cols)
-}
-
-#[cfg(test)]
-mod tests {
-    use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
-    use common_query::physical_plan::SessionContext;
-    use futures_util::StreamExt;
-    use table::table::numbers::NumbersTable;
-
-    use super::*;
-    use crate::local::memory::new_memory_catalog_list;
-    use crate::CatalogList;
-
-    #[tokio::test]
-    async fn test_tables() {
-        let catalog_list = new_memory_catalog_list().unwrap();
-        let schema = catalog_list
-            .catalog(DEFAULT_CATALOG_NAME)
-            .unwrap()
-            .unwrap()
-            .schema(DEFAULT_SCHEMA_NAME)
-            .unwrap()
-            .unwrap();
-        schema
-            .register_table("test_table".to_string(), Arc::new(NumbersTable::default()))
-            .unwrap();
-
-        let tables = Tables::new(catalog_list, "test_engine".to_string());
-        let tables_stream = tables.scan(None, &[], None).await.unwrap();
-        let session_ctx = SessionContext::new();
-        let mut tables_stream = tables_stream.execute(0, session_ctx.task_ctx()).unwrap();
-
-        if let Some(t) = tables_stream.next().await {
-            let batch = t.unwrap();
-            assert_eq!(1, batch.num_rows());
-            assert_eq!(4, batch.num_columns());
-            assert_eq!(
-                ConcreteDataType::string_datatype(),
-                batch.column(0).data_type()
-            );
-            assert_eq!(
-                ConcreteDataType::string_datatype(),
-                batch.column(1).data_type()
-            );
-            assert_eq!(
-                ConcreteDataType::string_datatype(),
-                batch.column(2).data_type()
-            );
-            assert_eq!(
-                ConcreteDataType::string_datatype(),
-                batch.column(3).data_type()
-            );
-            assert_eq!(
-                "greptime",
-                batch.column(0).get_ref(0).as_string().unwrap().unwrap()
-            );
-
-            assert_eq!(
-                "public",
-                batch.column(1).get_ref(0).as_string().unwrap().unwrap()
-            );
-
-            assert_eq!(
-                "test_table",
-                batch.column(2).get_ref(0).as_string().unwrap().unwrap()
-            );
-
-            assert_eq!(
-                "test_engine",
-                batch.column(3).get_ref(0).as_string().unwrap().unwrap()
-            );
-        } else {
-            panic!("Record batch should not be empty!")
-        }
-    }
-}
--- a/src/catalog/tests/local_catalog_tests.rs
+++ b/src/catalog/tests/local_catalog_tests.rs
@@ -21,6 +21,7 @@ mod tests {
    use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
    use common_telemetry::{error, info};
    use mito::config::EngineConfig;
+    use table::engine::manager::MemoryTableEngineManager;
    use table::table::numbers::NumbersTable;
    use table::TableRef;
    use tokio::sync::Mutex;
@@ -33,7 +34,8 @@ mod tests {
            mito::table::test_util::MockEngine::default(),
            object_store,
        ));
-        let catalog_manager = LocalCatalogManager::try_new(mock_engine).await.unwrap();
+        let engine_manager = Arc::new(MemoryTableEngineManager::new(mock_engine.clone()));
+        let catalog_manager = LocalCatalogManager::try_new(engine_manager).await.unwrap();
        catalog_manager.start().await?;
        Ok(catalog_manager)
    }
@@ -71,6 +73,7 @@ mod tests {

        let registered_table = catalog_manager
            .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, new_table_name)
+            .await
            .unwrap()
            .unwrap();
        assert_eq!(registered_table.table_info().ident.table_id, table_id);
@@ -158,6 +161,7 @@ mod tests {
            let table = guard.as_ref().unwrap();
            let table_registered = catalog_manager
                .table(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, "test_table")
+                .await
                .unwrap()
                .unwrap();
            assert_eq!(
--- a/src/catalog/tests/mock.rs
+++ b/src/catalog/tests/mock.rs
@@ -147,6 +147,7 @@ impl TableEngine for MockTableEngine {
        let table_id = TableId::from_str(
            request
                .table_options
+                .extra_options
                .get("table_id")
                .unwrap_or(&default_table_id),
        )
@@ -220,4 +221,8 @@ impl TableEngine for MockTableEngine {
    ) -> table::Result<bool> {
        unimplemented!()
    }
+
+    async fn close(&self) -> table::Result<()> {
+        Ok(())
+    }
 }
--- a/src/catalog/tests/remote_catalog_tests.rs
+++ b/src/catalog/tests/remote_catalog_tests.rs
@@ -27,9 +27,10 @@ mod tests {
        KvBackend, KvBackendRef, RemoteCatalogManager, RemoteCatalogProvider, RemoteSchemaProvider,
    };
    use catalog::{CatalogList, CatalogManager, RegisterTableRequest};
-    use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
+    use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, MITO_ENGINE};
    use datatypes::schema::RawSchema;
    use futures_util::StreamExt;
+    use table::engine::manager::MemoryTableEngineManager;
    use table::engine::{EngineContext, TableEngineRef};
    use table::requests::CreateTableRequest;

@@ -80,8 +81,11 @@ mod tests {
    ) -> (KvBackendRef, TableEngineRef, Arc<RemoteCatalogManager>) {
        let backend = Arc::new(MockKvBackend::default()) as KvBackendRef;
        let table_engine = Arc::new(MockTableEngine::default());
-        let catalog_manager =
-            RemoteCatalogManager::new(table_engine.clone(), node_id, backend.clone());
+        let engine_manager = Arc::new(MemoryTableEngineManager::alias(
+            MITO_ENGINE.to_string(),
+            table_engine.clone(),
+        ));
+        let catalog_manager = RemoteCatalogManager::new(engine_manager, node_id, backend.clone());
        catalog_manager.start().await.unwrap();
        (backend, table_engine, Arc::new(catalog_manager))
    }
@@ -131,6 +135,7 @@ mod tests {
                    primary_key_indices: vec![],
                    create_if_not_exists: false,
                    table_options: Default::default(),
+                    engine: MITO_ENGINE.to_string(),
                },
            )
            .await
@@ -168,7 +173,6 @@ mod tests {
            .schema(DEFAULT_SCHEMA_NAME)
            .unwrap()
            .unwrap();
-        assert_eq!(vec!["numbers"], default_schema.table_names().unwrap());

        // register a new table with an nonexistent catalog
        let catalog_name = DEFAULT_CATALOG_NAME.to_string();
@@ -191,6 +195,7 @@ mod tests {
                    primary_key_indices: vec![],
                    create_if_not_exists: false,
                    table_options: Default::default(),
+                    engine: MITO_ENGINE.to_string(),
                },
            )
            .await
@@ -203,14 +208,7 @@ mod tests {
            table,
        };
        assert!(catalog_manager.register_table(reg_req).await.unwrap());
-        assert_eq!(
-            HashSet::from([table_name, "numbers".to_string()]),
-            default_schema
-                .table_names()
-                .unwrap()
-                .into_iter()
-                .collect::<HashSet<_>>()
-        );
+        assert_eq!(vec![table_name], default_schema.table_names().unwrap());
    }

    #[tokio::test]
@@ -251,6 +249,7 @@ mod tests {
                    primary_key_indices: vec![],
                    create_if_not_exists: false,
                    table_options: Default::default(),
+                    engine: MITO_ENGINE.to_string(),
                },
            )
            .await
--- a/src/client/Cargo.toml
+++ b/src/client/Cargo.toml
@@ -16,13 +16,14 @@ common-grpc-expr = { path = "../common/grpc-expr" }
 common-query = { path = "../common/query" }
 common-recordbatch = { path = "../common/recordbatch" }
 common-time = { path = "../common/time" }
+common-telemetry = { path = "../common/telemetry" }
 datafusion.workspace = true
 datatypes = { path = "../datatypes" }
 enum_dispatch = "0.3"
 futures-util.workspace = true
 parking_lot = "0.12"
 prost.workspace = true
-rand = "0.8"
+rand.workspace = true
 snafu.workspace = true
 tonic.workspace = true

@@ -32,12 +33,8 @@ substrait = { path = "../common/substrait" }
 tokio.workspace = true
 tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
-
-# TODO(ruihang): upgrade to 0.11 once substrait-rs supports it.
-[dev-dependencies.prost_09]
-package = "prost"
-version = "0.9"
+prost.workspace = true

 [dev-dependencies.substrait_proto]
 package = "substrait"
-version = "0.2"
+version = "0.4"
--- a/src/client/examples/logical.rs
+++ b/src/client/examples/logical.rs
@@ -14,11 +14,12 @@

 use api::v1::{ColumnDataType, ColumnDef, CreateTableExpr, TableId};
 use client::{Client, Database};
-use prost_09::Message;
-use substrait_proto::protobuf::plan_rel::RelType as PlanRelType;
-use substrait_proto::protobuf::read_rel::{NamedTable, ReadType};
-use substrait_proto::protobuf::rel::RelType;
-use substrait_proto::protobuf::{PlanRel, ReadRel, Rel};
+use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, MITO_ENGINE};
+use prost::Message;
+use substrait_proto::proto::plan_rel::RelType as PlanRelType;
+use substrait_proto::proto::read_rel::{NamedTable, ReadType};
+use substrait_proto::proto::rel::RelType;
+use substrait_proto::proto::{PlanRel, ReadRel, Rel};
 use tracing::{event, Level};

 fn main() {
@@ -63,9 +64,10 @@ async fn run() {
        table_options: Default::default(),
        table_id: Some(TableId { id: 1024 }),
        region_ids: vec![0],
+        engine: MITO_ENGINE.to_string(),
    };

-    let db = Database::with_client(client);
+    let db = Database::new(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, client);
    let result = db.create(create_table_expr).await.unwrap();
    event!(Level::INFO, "create table result: {:#?}", result);

@@ -88,12 +90,8 @@ fn mock_logical_plan() -> Vec<u8> {
    let read_type = ReadType::NamedTable(named_table);

    let read_rel = ReadRel {
-        common: None,
-        base_schema: None,
-        filter: None,
-        projection: None,
-        advanced_extension: None,
        read_type: Some(read_type),
+        ..Default::default()
    };

    let mut buf = vec![];
--- a/src/client/src/client.rs
+++ b/src/client/src/client.rs
@@ -14,6 +14,9 @@

 use std::sync::Arc;

+use api::v1::greptime_database_client::GreptimeDatabaseClient;
+use api::v1::health_check_client::HealthCheckClient;
+use api::v1::HealthCheckRequest;
 use arrow_flight::flight_service_client::FlightServiceClient;
 use common_grpc::channel_manager::ChannelManager;
 use parking_lot::RwLock;
@@ -23,6 +26,10 @@ use tonic::transport::Channel;
 use crate::load_balance::{LoadBalance, Loadbalancer};
 use crate::{error, Result};

+pub(crate) struct DatabaseClient {
+    pub(crate) inner: GreptimeDatabaseClient<Channel>,
+}
+
 pub(crate) struct FlightClient {
    addr: String,
    client: FlightServiceClient<Channel>,
@@ -118,7 +125,7 @@ impl Client {
        self.inner.set_peers(urls);
    }

-    pub(crate) fn make_client(&self) -> Result<FlightClient> {
+    fn find_channel(&self) -> Result<(String, Channel)> {
        let addr = self
            .inner
            .get_peer()
@@ -131,11 +138,30 @@ impl Client {
            .channel_manager
            .get(&addr)
            .context(error::CreateChannelSnafu { addr: &addr })?;
+        Ok((addr, channel))
+    }
+
+    pub(crate) fn make_flight_client(&self) -> Result<FlightClient> {
+        let (addr, channel) = self.find_channel()?;
        Ok(FlightClient {
            addr,
            client: FlightServiceClient::new(channel),
        })
    }
+
+    pub(crate) fn make_database_client(&self) -> Result<DatabaseClient> {
+        let (_, channel) = self.find_channel()?;
+        Ok(DatabaseClient {
+            inner: GreptimeDatabaseClient::new(channel),
+        })
+    }
+
+    pub async fn health_check(&self) -> Result<()> {
+        let (_, channel) = self.find_channel()?;
+        let mut client = HealthCheckClient::new(channel);
+        client.health_check(HealthCheckRequest {}).await?;
+        Ok(())
+    }
 }

 #[cfg(test)]
--- a/src/client/src/database.rs
+++ b/src/client/src/database.rs
@@ -12,57 +12,121 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-use std::str::FromStr;
-
+use api::v1::auth_header::AuthScheme;
 use api::v1::ddl_request::Expr as DdlExpr;
 use api::v1::greptime_request::Request;
 use api::v1::query_request::Query;
 use api::v1::{
-    AlterExpr, CreateTableExpr, DdlRequest, DropTableExpr, GreptimeRequest, InsertRequest,
-    QueryRequest, RequestHeader,
+    greptime_response, AffectedRows, AlterExpr, AuthHeader, CreateTableExpr, DdlRequest,
+    DropTableExpr, FlushTableExpr, GreptimeRequest, InsertRequest, PromRangeQuery, QueryRequest,
+    RequestHeader,
 };
 use arrow_flight::{FlightData, Ticket};
-use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
 use common_error::prelude::*;
 use common_grpc::flight::{flight_messages_to_recordbatches, FlightDecoder, FlightMessage};
 use common_query::Output;
+use common_telemetry::logging;
 use futures_util::{TryFutureExt, TryStreamExt};
 use prost::Message;
 use snafu::{ensure, ResultExt};

-use crate::error::{ConvertFlightDataSnafu, IllegalFlightMessagesSnafu};
+use crate::error::{
+    ConvertFlightDataSnafu, IllegalDatabaseResponseSnafu, IllegalFlightMessagesSnafu,
+};
 use crate::{error, Client, Result};

-#[derive(Clone, Debug)]
+#[derive(Clone, Debug, Default)]
 pub struct Database {
    // The "catalog" and "schema" to be used in processing the requests at the server side.
    // They are the "hint" or "context", just like how the "database" in "USE" statement is treated in MySQL.
    // They will be carried in the request header.
    catalog: String,
    schema: String,
+    // The dbname follows naming rule as out mysql, postgres and http
+    // protocol. The server treat dbname in priority of catalog/schema.
+    dbname: String,

    client: Client,
+    ctx: FlightContext,
 }

 impl Database {
+    /// Create database service client using catalog and schema
    pub fn new(catalog: impl Into<String>, schema: impl Into<String>, client: Client) -> Self {
        Self {
            catalog: catalog.into(),
            schema: schema.into(),
            client,
+            ..Default::default()
        }
    }

-    pub fn with_client(client: Client) -> Self {
-        Self::new(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, client)
+    /// Create database service client using dbname.
+    ///
+    /// This API is designed for external usage. `dbname` is:
+    ///
+    /// - the name of database when using GreptimeDB standalone or cluster
+    /// - the name provided by GreptimeCloud or other multi-tenant GreptimeDB
+    /// environment
+    pub fn new_with_dbname(dbname: impl Into<String>, client: Client) -> Self {
+        Self {
+            dbname: dbname.into(),
+            client,
+            ..Default::default()
+        }
+    }
+
+    pub fn catalog(&self) -> &String {
+        &self.catalog
+    }
+
+    pub fn set_catalog(&mut self, catalog: impl Into<String>) {
+        self.catalog = catalog.into();
+    }
+
+    pub fn schema(&self) -> &String {
+        &self.schema
    }

    pub fn set_schema(&mut self, schema: impl Into<String>) {
        self.schema = schema.into();
    }

-    pub async fn insert(&self, request: InsertRequest) -> Result<Output> {
-        self.do_get(Request::Insert(request)).await
+    pub fn dbname(&self) -> &String {
+        &self.dbname
+    }
+
+    pub fn set_dbname(&mut self, dbname: impl Into<String>) {
+        self.dbname = dbname.into();
+    }
+
+    pub fn set_auth(&mut self, auth: AuthScheme) {
+        self.ctx.auth_header = Some(AuthHeader {
+            auth_scheme: Some(auth),
+        });
+    }
+
+    pub async fn insert(&self, request: InsertRequest) -> Result<u32> {
+        let mut client = self.client.make_database_client()?.inner;
+        let request = GreptimeRequest {
+            header: Some(RequestHeader {
+                catalog: self.catalog.clone(),
+                schema: self.schema.clone(),
+                authorization: self.ctx.auth_header.clone(),
+                dbname: self.dbname.clone(),
+            }),
+            request: Some(Request::Insert(request)),
+        };
+        let response = client
+            .handle(request)
+            .await?
+            .into_inner()
+            .response
+            .context(IllegalDatabaseResponseSnafu {
+                err_msg: "GreptimeResponse is empty",
+            })?;
+        let greptime_response::Response::AffectedRows(AffectedRows { value }) = response;
+        Ok(value)
    }

    pub async fn sql(&self, sql: &str) -> Result<Output> {
@@ -79,6 +143,24 @@ impl Database {
        .await
    }

+    pub async fn prom_range_query(
+        &self,
+        promql: &str,
+        start: &str,
+        end: &str,
+        step: &str,
+    ) -> Result<Output> {
+        self.do_get(Request::Query(QueryRequest {
+            query: Some(Query::PromRangeQuery(PromRangeQuery {
+                query: promql.to_string(),
+                start: start.to_string(),
+                end: end.to_string(),
+                step: step.to_string(),
+            })),
+        }))
+        .await
+    }
+
    pub async fn create(&self, expr: CreateTableExpr) -> Result<Output> {
        self.do_get(Request::Ddl(DdlRequest {
            expr: Some(DdlExpr::CreateTable(expr)),
@@ -100,19 +182,28 @@ impl Database {
        .await
    }

+    pub async fn flush_table(&self, expr: FlushTableExpr) -> Result<Output> {
+        self.do_get(Request::Ddl(DdlRequest {
+            expr: Some(DdlExpr::FlushTable(expr)),
+        }))
+        .await
+    }
+
    async fn do_get(&self, request: Request) -> Result<Output> {
        let request = GreptimeRequest {
            header: Some(RequestHeader {
                catalog: self.catalog.clone(),
                schema: self.schema.clone(),
+                authorization: self.ctx.auth_header.clone(),
+                dbname: self.dbname.clone(),
            }),
            request: Some(request),
        };
        let request = Ticket {
-            ticket: request.encode_to_vec(),
+            ticket: request.encode_to_vec().into(),
        };

-        let mut client = self.client.make_client()?;
+        let mut client = self.client.make_flight_client()?;

        // TODO(LFC): Streaming get flight data.
        let flight_data: Vec<FlightData> = client
@@ -121,17 +212,26 @@ impl Database {
            .and_then(|response| response.into_inner().try_collect())
            .await
            .map_err(|e| {
-                let code = get_metadata_value(&e, INNER_ERROR_CODE)
-                    .and_then(|s| StatusCode::from_str(&s).ok())
-                    .unwrap_or(StatusCode::Unknown);
-                let msg = get_metadata_value(&e, INNER_ERROR_MSG).unwrap_or(e.to_string());
-                error::ExternalSnafu { code, msg }
+                let tonic_code = e.code();
+                let e: error::Error = e.into();
+                let code = e.status_code();
+                let msg = e.to_string();
+                error::ServerSnafu { code, msg }
                    .fail::<()>()
                    .map_err(BoxedError::new)
                    .context(error::FlightGetSnafu {
-                        tonic_code: e.code(),
+                        tonic_code,
                        addr: client.addr(),
                    })
+                    .map_err(|error| {
+                        logging::error!(
+                            "Failed to do Flight get, addr: {}, code: {}, source: {}",
+                            client.addr(),
+                            tonic_code,
+                            error
+                        );
+                        error
+                    })
                    .unwrap_err()
            })?;

@@ -158,10 +258,9 @@ impl Database {
    }
 }

-fn get_metadata_value(e: &tonic::Status, key: &str) -> Option<String> {
-    e.metadata()
-        .get(key)
-        .and_then(|v| String::from_utf8(v.as_bytes().to_vec()).ok())
+#[derive(Default, Debug, Clone)]
+pub struct FlightContext {
+    auth_header: Option<AuthHeader>,
 }

 #[cfg(test)]
@@ -169,7 +268,8 @@ mod tests {
    use std::sync::Arc;

    use api::helper::ColumnDataTypeWrapper;
-    use api::v1::Column;
+    use api::v1::auth_header::AuthScheme;
+    use api::v1::{AuthHeader, Basic, Column};
    use common_grpc::select::{null_mask, values};
    use common_grpc_expr::column_to_vector;
    use datatypes::prelude::{Vector, VectorRef};
@@ -179,6 +279,8 @@ mod tests {
        UInt32Vector, UInt64Vector, UInt8Vector,
    };

+    use crate::database::FlightContext;
+
    #[test]
    fn test_column_to_vector() {
        let mut column = create_test_column(Arc::new(BooleanVector::from(vec![true])));
@@ -262,4 +364,26 @@ mod tests {
            datatype: wrapper.datatype() as i32,
        }
    }
+
+    #[test]
+    fn test_flight_ctx() {
+        let mut ctx = FlightContext::default();
+        assert!(ctx.auth_header.is_none());
+
+        let basic = AuthScheme::Basic(Basic {
+            username: "u".to_string(),
+            password: "p".to_string(),
+        });
+
+        ctx.auth_header = Some(AuthHeader {
+            auth_scheme: Some(basic),
+        });
+
+        assert!(matches!(
+            ctx.auth_header,
+            Some(AuthHeader {
+                auth_scheme: Some(AuthScheme::Basic(_)),
+            })
+        ))
+    }
 }
--- a/src/client/src/error.rs
+++ b/src/client/src/error.rs
@@ -13,25 +13,19 @@
 // limitations under the License.

 use std::any::Any;
+use std::str::FromStr;

 use common_error::prelude::*;
-use tonic::Code;
+use snafu::Location;
+use tonic::{Code, Status};

 #[derive(Debug, Snafu)]
 #[snafu(visibility(pub))]
 pub enum Error {
    #[snafu(display("Illegal Flight messages, reason: {}", reason))]
-    IllegalFlightMessages {
-        reason: String,
-        backtrace: Backtrace,
-    },
+    IllegalFlightMessages { reason: String, location: Location },

-    #[snafu(display(
-        "Failed to do Flight get, addr: {}, code: {}, source: {}",
-        addr,
-        tonic_code,
-        source
-    ))]
+    #[snafu(display("Failed to do Flight get, code: {}, source: {}", tonic_code, source))]
    FlightGet {
        addr: String,
        tonic_code: Code,
@@ -51,13 +45,10 @@ pub enum Error {
    },

    #[snafu(display("Illegal GRPC client state: {}", err_msg))]
-    IllegalGrpcClientState {
-        err_msg: String,
-        backtrace: Backtrace,
-    },
+    IllegalGrpcClientState { err_msg: String, location: Location },

    #[snafu(display("Missing required field in protobuf, field: {}", field))]
-    MissingField { field: String, backtrace: Backtrace },
+    MissingField { field: String, location: Location },

    #[snafu(display(
        "Failed to create gRPC channel, peer address: {}, source: {}",
@@ -70,9 +61,12 @@ pub enum Error {
        source: common_grpc::error::Error,
    },

-    /// Error deserialized from gRPC metadata
+    // Server error carried in Tonic Status's metadata.
    #[snafu(display("{}", msg))]
-    ExternalError { code: StatusCode, msg: String },
+    Server { code: StatusCode, msg: String },
+
+    #[snafu(display("Illegal Database response: {err_msg}"))]
+    IllegalDatabaseResponse { err_msg: String },
 }

 pub type Result<T> = std::result::Result<T, Error>;
@@ -82,21 +76,37 @@ impl ErrorExt for Error {
        match self {
            Error::IllegalFlightMessages { .. }
            | Error::ColumnDataType { .. }
-            | Error::MissingField { .. } => StatusCode::Internal,
+            | Error::MissingField { .. }
+            | Error::IllegalDatabaseResponse { .. } => StatusCode::Internal,
+
+            Error::Server { code, .. } => *code,
            Error::FlightGet { source, .. } => source.status_code(),
            Error::CreateChannel { source, .. } | Error::ConvertFlightData { source } => {
                source.status_code()
            }
            Error::IllegalGrpcClientState { .. } => StatusCode::Unexpected,
-            Error::ExternalError { code, .. } => *code,
        }
    }

-    fn backtrace_opt(&self) -> Option<&Backtrace> {
-        ErrorCompat::backtrace(self)
-    }
-
    fn as_any(&self) -> &dyn Any {
        self
    }
 }
+
+impl From<Status> for Error {
+    fn from(e: Status) -> Self {
+        fn get_metadata_value(e: &Status, key: &str) -> Option<String> {
+            e.metadata()
+                .get(key)
+                .and_then(|v| String::from_utf8(v.as_bytes().to_vec()).ok())
+        }
+
+        let code = get_metadata_value(&e, INNER_ERROR_CODE)
+            .and_then(|s| StatusCode::from_str(&s).ok())
+            .unwrap_or(StatusCode::Unknown);
+
+        let msg = get_metadata_value(&e, INNER_ERROR_MSG).unwrap_or(e.to_string());
+
+        Self::Server { code, msg }
+    }
+}
--- a/src/client/src/lib.rs
+++ b/src/client/src/lib.rs
@@ -18,6 +18,7 @@ mod error;
 pub mod load_balance;

 pub use api;
+pub use common_catalog::consts::{DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};

 pub use self::client::Client;
 pub use self::database::Database;
--- a/src/cmd/Cargo.toml
+++ b/src/cmd/Cargo.toml
@@ -9,28 +9,46 @@ default-run = "greptime"
 name = "greptime"
 path = "src/bin/greptime.rs"

+[features]
+mem-prof = ["tikv-jemallocator", "tikv-jemalloc-ctl"]
+
 [dependencies]
 anymap = "1.0.0-beta.2"
+catalog = { path = "../catalog" }
 clap = { version = "3.1", features = ["derive"] }
+client = { path = "../client" }
 common-base = { path = "../common/base" }
 common-error = { path = "../common/error" }
+common-query = { path = "../common/query" }
+common-recordbatch = { path = "../common/recordbatch" }
 common-telemetry = { path = "../common/telemetry", features = [
    "deadlock_detection",
 ] }
 datanode = { path = "../datanode" }
+either = "1.8"
 frontend = { path = "../frontend" }
 futures.workspace = true
 meta-client = { path = "../meta-client" }
 meta-srv = { path = "../meta-srv" }
+nu-ansi-term = "0.46"
+partition = { path = "../partition" }
+query = { path = "../query" }
+rustyline = "10.1"
 serde.workspace = true
 servers = { path = "../servers" }
+session = { path = "../session" }
 snafu.workspace = true
+substrait = { path = "../common/substrait" }
+tikv-jemalloc-ctl = { version = "0.5", optional = true }
+tikv-jemallocator = { version = "0.5", optional = true }
 tokio.workspace = true
 toml = "0.5"

+
 [dev-dependencies]
+common-test-util = { path = "../common/test-util" }
+rexpect = "0.5"
 serde.workspace = true
-tempdir = "0.3"

 [build-dependencies]
 build-data = "0.1.3"
--- a/src/cmd/src/bin/greptime.rs
+++ b/src/cmd/src/bin/greptime.rs
@@ -12,11 +12,13 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

+#![doc = include_str!("../../../../README.md")]
+
 use std::fmt;

 use clap::Parser;
 use cmd::error::Result;
-use cmd::{datanode, frontend, metasrv, standalone};
+use cmd::{cli, datanode, frontend, metasrv, standalone};
 use common_telemetry::logging::{error, info};

 #[derive(Parser)]
@@ -30,9 +32,39 @@ struct Command {
    subcmd: SubCommand,
 }

+pub enum Application {
+    Datanode(datanode::Instance),
+    Frontend(frontend::Instance),
+    Metasrv(metasrv::Instance),
+    Standalone(standalone::Instance),
+    Cli(cli::Instance),
+}
+
+impl Application {
+    async fn run(&mut self) -> Result<()> {
+        match self {
+            Application::Datanode(instance) => instance.run().await,
+            Application::Frontend(instance) => instance.run().await,
+            Application::Metasrv(instance) => instance.run().await,
+            Application::Standalone(instance) => instance.run().await,
+            Application::Cli(instance) => instance.run().await,
+        }
+    }
+
+    async fn stop(&self) -> Result<()> {
+        match self {
+            Application::Datanode(instance) => instance.stop().await,
+            Application::Frontend(instance) => instance.stop().await,
+            Application::Metasrv(instance) => instance.stop().await,
+            Application::Standalone(instance) => instance.stop().await,
+            Application::Cli(instance) => instance.stop().await,
+        }
+    }
+}
+
 impl Command {
-    async fn run(self) -> Result<()> {
-        self.subcmd.run().await
+    async fn build(self) -> Result<Application> {
+        self.subcmd.build().await
    }
 }

@@ -46,15 +78,33 @@ enum SubCommand {
    Metasrv(metasrv::Command),
    #[clap(name = "standalone")]
    Standalone(standalone::Command),
+    #[clap(name = "cli")]
+    Cli(cli::Command),
 }

 impl SubCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Application> {
        match self {
-            SubCommand::Datanode(cmd) => cmd.run().await,
-            SubCommand::Frontend(cmd) => cmd.run().await,
-            SubCommand::Metasrv(cmd) => cmd.run().await,
-            SubCommand::Standalone(cmd) => cmd.run().await,
+            SubCommand::Datanode(cmd) => {
+                let app = cmd.build().await?;
+                Ok(Application::Datanode(app))
+            }
+            SubCommand::Frontend(cmd) => {
+                let app = cmd.build().await?;
+                Ok(Application::Frontend(app))
+            }
+            SubCommand::Metasrv(cmd) => {
+                let app = cmd.build().await?;
+                Ok(Application::Metasrv(app))
+            }
+            SubCommand::Standalone(cmd) => {
+                let app = cmd.build().await?;
+                Ok(Application::Standalone(app))
+            }
+            SubCommand::Cli(cmd) => {
+                let app = cmd.build().await?;
+                Ok(Application::Cli(app))
+            }
        }
    }
 }
@@ -66,6 +116,7 @@ impl fmt::Display for SubCommand {
            SubCommand::Frontend(..) => write!(f, "greptime-frontend"),
            SubCommand::Metasrv(..) => write!(f, "greptime-metasrv"),
            SubCommand::Standalone(..) => write!(f, "greptime-standalone"),
+            SubCommand::Cli(_) => write!(f, "greptime-cli"),
        }
    }
 }
@@ -83,6 +134,10 @@ fn print_version() -> &'static str {
    )
 }

+#[cfg(feature = "mem-prof")]
+#[global_allocator]
+static ALLOC: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
+
 #[tokio::main]
 async fn main() -> Result<()> {
    let cmd = Command::parse();
@@ -96,13 +151,18 @@ async fn main() -> Result<()> {
    common_telemetry::init_default_metrics_recorder();
    let _guard = common_telemetry::init_global_logging(app_name, log_dir, log_level, false);

+    let mut app = cmd.build().await?;
+
    tokio::select! {
-        result = cmd.run() => {
+        result = app.run() => {
            if let Err(err) = result {
                error!(err; "Fatal error occurs!");
            }
        }
        _ = tokio::signal::ctrl_c() => {
+            if let Err(err) = app.stop().await {
+                error!(err; "Fatal error occurs!");
+            }
            info!("Goodbye!");
        }
    }
--- a/src/cmd/src/cli.rs
+++ b/src/cmd/src/cli.rs
@@ -0,0 +1,78 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+mod cmd;
+mod helper;
+mod repl;
+
+use clap::Parser;
+pub use repl::Repl;
+
+use crate::error::Result;
+
+pub struct Instance {
+    repl: Repl,
+}
+
+impl Instance {
+    pub async fn run(&mut self) -> Result<()> {
+        self.repl.run().await
+    }
+
+    pub async fn stop(&self) -> Result<()> {
+        Ok(())
+    }
+}
+
+#[derive(Parser)]
+pub struct Command {
+    #[clap(subcommand)]
+    cmd: SubCommand,
+}
+
+impl Command {
+    pub async fn build(self) -> Result<Instance> {
+        self.cmd.build().await
+    }
+}
+
+#[derive(Parser)]
+enum SubCommand {
+    Attach(AttachCommand),
+}
+
+impl SubCommand {
+    async fn build(self) -> Result<Instance> {
+        match self {
+            SubCommand::Attach(cmd) => cmd.build().await,
+        }
+    }
+}
+
+#[derive(Debug, Parser)]
+pub(crate) struct AttachCommand {
+    #[clap(long)]
+    pub(crate) grpc_addr: String,
+    #[clap(long)]
+    pub(crate) meta_addr: Option<String>,
+    #[clap(long, action)]
+    pub(crate) disable_helper: bool,
+}
+
+impl AttachCommand {
+    async fn build(self) -> Result<Instance> {
+        let repl = Repl::try_new(&self).await?;
+        Ok(Instance { repl })
+    }
+}
--- a/src/cmd/src/cli/cmd.rs
+++ b/src/cmd/src/cli/cmd.rs
@@ -0,0 +1,154 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use crate::error::{Error, InvalidReplCommandSnafu, Result};
+
+/// Represents the parsed command from the user (which may be over many lines)
+#[derive(Debug, PartialEq)]
+pub(crate) enum ReplCommand {
+    Help,
+    UseDatabase { db_name: String },
+    Sql { sql: String },
+    Exit,
+}
+
+impl TryFrom<&str> for ReplCommand {
+    type Error = Error;
+
+    fn try_from(input: &str) -> Result<Self> {
+        let input = input.trim();
+        if input.is_empty() {
+            return InvalidReplCommandSnafu {
+                reason: "No command specified".to_string(),
+            }
+            .fail();
+        }
+
+        // If line ends with ';', it must be treated as a complete input.
+        // However, the opposite is not true.
+        let input_is_completed = input.ends_with(';');
+
+        let input = input.strip_suffix(';').map(|x| x.trim()).unwrap_or(input);
+        let lowercase = input.to_lowercase();
+        match lowercase.as_str() {
+            "help" => Ok(Self::Help),
+            "exit" | "quit" => Ok(Self::Exit),
+            _ => match input.split_once(' ') {
+                Some((maybe_use, database)) if maybe_use.to_lowercase() == "use" => {
+                    Ok(Self::UseDatabase {
+                        db_name: database.trim().to_string(),
+                    })
+                }
+                // Any valid SQL must contains at least one whitespace.
+                Some(_) if input_is_completed => Ok(Self::Sql {
+                    sql: input.to_string(),
+                }),
+                _ => InvalidReplCommandSnafu {
+                    reason: format!("unknown command '{input}', maybe input is not completed"),
+                }
+                .fail(),
+            },
+        }
+    }
+}
+
+impl ReplCommand {
+    pub fn help() -> &'static str {
+        r#"
+Available commands (case insensitive):
+- 'help': print this help
+- 'exit' or 'quit': exit the REPL
+- 'use <your database name>': switch to another database/schema context
+- Other typed in text will be treated as SQL.
+  You can enter new line while typing, just remember to end it with ';'.
+"#
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::error::Error::InvalidReplCommand;
+
+    #[test]
+    fn test_from_str() {
+        fn test_ok(s: &str, expected: ReplCommand) {
+            let actual: ReplCommand = s.try_into().unwrap();
+            assert_eq!(expected, actual, "'{}'", s);
+        }
+
+        fn test_err(s: &str) {
+            let result: Result<ReplCommand> = s.try_into();
+            assert!(matches!(result, Err(InvalidReplCommand { .. })))
+        }
+
+        test_err("");
+        test_err("  ");
+        test_err("\t");
+
+        test_ok("help", ReplCommand::Help);
+        test_ok("help", ReplCommand::Help);
+        test_ok("  help", ReplCommand::Help);
+        test_ok("  help  ", ReplCommand::Help);
+        test_ok("  HELP  ", ReplCommand::Help);
+        test_ok("  Help;  ", ReplCommand::Help);
+        test_ok("  help  ; ", ReplCommand::Help);
+
+        test_ok("exit", ReplCommand::Exit);
+        test_ok("exit;", ReplCommand::Exit);
+        test_ok("exit ;", ReplCommand::Exit);
+        test_ok("EXIT", ReplCommand::Exit);
+
+        test_ok("quit", ReplCommand::Exit);
+        test_ok("quit;", ReplCommand::Exit);
+        test_ok("quit ;", ReplCommand::Exit);
+        test_ok("QUIT", ReplCommand::Exit);
+
+        test_ok(
+            "use Foo",
+            ReplCommand::UseDatabase {
+                db_name: "Foo".to_string(),
+            },
+        );
+        test_ok(
+            "  use Foo ;  ",
+            ReplCommand::UseDatabase {
+                db_name: "Foo".to_string(),
+            },
+        );
+        // ensure that database name is case sensitive
+        test_ok(
+            "  use FOO ;  ",
+            ReplCommand::UseDatabase {
+                db_name: "FOO".to_string(),
+            },
+        );
+
+        // ensure that we aren't messing with capitalization
+        test_ok(
+            "SELECT * from foo;",
+            ReplCommand::Sql {
+                sql: "SELECT * from foo".to_string(),
+            },
+        );
+        // Input line (that don't belong to any other cases above) must ends with ';' to make it a valid SQL.
+        test_err("insert blah");
+        test_ok(
+            "insert blah;",
+            ReplCommand::Sql {
+                sql: "insert blah".to_string(),
+            },
+        );
+    }
+}
--- a/src/cmd/src/cli/helper.rs
+++ b/src/cmd/src/cli/helper.rs
@@ -0,0 +1,112 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::borrow::Cow;
+
+use rustyline::completion::Completer;
+use rustyline::highlight::{Highlighter, MatchingBracketHighlighter};
+use rustyline::hint::{Hinter, HistoryHinter};
+use rustyline::validate::{ValidationContext, ValidationResult, Validator};
+
+use crate::cli::cmd::ReplCommand;
+
+pub(crate) struct RustylineHelper {
+    hinter: HistoryHinter,
+    highlighter: MatchingBracketHighlighter,
+}
+
+impl Default for RustylineHelper {
+    fn default() -> Self {
+        Self {
+            hinter: HistoryHinter {},
+            highlighter: MatchingBracketHighlighter::default(),
+        }
+    }
+}
+
+impl rustyline::Helper for RustylineHelper {}
+
+impl Validator for RustylineHelper {
+    fn validate(&self, ctx: &mut ValidationContext<'_>) -> rustyline::Result<ValidationResult> {
+        let input = ctx.input();
+        match ReplCommand::try_from(input) {
+            Ok(_) => Ok(ValidationResult::Valid(None)),
+            Err(e) => {
+                if input.trim_end().ends_with(';') {
+                    // If line ends with ';', it HAS to be a valid command.
+                    Ok(ValidationResult::Invalid(Some(e.to_string())))
+                } else {
+                    Ok(ValidationResult::Incomplete)
+                }
+            }
+        }
+    }
+}
+
+impl Hinter for RustylineHelper {
+    type Hint = String;
+
+    fn hint(&self, line: &str, pos: usize, ctx: &rustyline::Context<'_>) -> Option<Self::Hint> {
+        self.hinter.hint(line, pos, ctx)
+    }
+}
+
+impl Highlighter for RustylineHelper {
+    fn highlight<'l>(&self, line: &'l str, pos: usize) -> Cow<'l, str> {
+        self.highlighter.highlight(line, pos)
+    }
+
+    fn highlight_prompt<'b, 's: 'b, 'p: 'b>(
+        &'s self,
+        prompt: &'p str,
+        default: bool,
+    ) -> Cow<'b, str> {
+        self.highlighter.highlight_prompt(prompt, default)
+    }
+
+    fn highlight_hint<'h>(&self, hint: &'h str) -> Cow<'h, str> {
+        use nu_ansi_term::Style;
+        Cow::Owned(Style::new().dimmed().paint(hint).to_string())
+    }
+
+    fn highlight_candidate<'c>(
+        &self,
+        candidate: &'c str,
+        completion: rustyline::CompletionType,
+    ) -> Cow<'c, str> {
+        self.highlighter.highlight_candidate(candidate, completion)
+    }
+
+    fn highlight_char(&self, line: &str, pos: usize) -> bool {
+        self.highlighter.highlight_char(line, pos)
+    }
+}
+
+impl Completer for RustylineHelper {
+    type Candidate = String;
+
+    fn complete(
+        &self,
+        line: &str,
+        pos: usize,
+        ctx: &rustyline::Context<'_>,
+    ) -> rustyline::Result<(usize, Vec<Self::Candidate>)> {
+        // If there is a hint, use that as the auto-complete when user hits `tab`
+        if let Some(hint) = self.hinter.hint(line, pos, ctx) {
+            Ok((pos, vec![hint]))
+        } else {
+            Ok((0, vec![]))
+        }
+    }
+}
--- a/src/cmd/src/cli/repl.rs
+++ b/src/cmd/src/cli/repl.rs
@@ -0,0 +1,273 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::path::PathBuf;
+use std::sync::Arc;
+use std::time::Instant;
+
+use catalog::remote::MetaKvBackend;
+use client::{Client, Database, DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME};
+use common_error::prelude::ErrorExt;
+use common_query::Output;
+use common_recordbatch::RecordBatches;
+use common_telemetry::logging;
+use either::Either;
+use frontend::catalog::FrontendCatalogManager;
+use frontend::datanode::DatanodeClients;
+use meta_client::client::MetaClientBuilder;
+use partition::manager::PartitionRuleManager;
+use partition::route::TableRoutes;
+use query::datafusion::DatafusionQueryEngine;
+use query::logical_optimizer::LogicalOptimizer;
+use query::parser::QueryLanguageParser;
+use query::plan::LogicalPlan;
+use query::query_engine::QueryEngineState;
+use query::QueryEngine;
+use rustyline::error::ReadlineError;
+use rustyline::Editor;
+use session::context::QueryContext;
+use snafu::{ErrorCompat, ResultExt};
+use substrait::{DFLogicalSubstraitConvertor, SubstraitPlan};
+
+use crate::cli::cmd::ReplCommand;
+use crate::cli::helper::RustylineHelper;
+use crate::cli::AttachCommand;
+use crate::error::{
+    CollectRecordBatchesSnafu, ParseSqlSnafu, PlanStatementSnafu, PrettyPrintRecordBatchesSnafu,
+    ReadlineSnafu, ReplCreationSnafu, RequestDatabaseSnafu, Result, StartMetaClientSnafu,
+    SubstraitEncodeLogicalPlanSnafu,
+};
+
+/// Captures the state of the repl, gathers commands and executes them one by one
+pub struct Repl {
+    /// Rustyline editor for interacting with user on command line
+    rl: Editor<RustylineHelper>,
+
+    /// Current prompt
+    prompt: String,
+
+    /// Client for interacting with GreptimeDB
+    database: Database,
+
+    query_engine: Option<DatafusionQueryEngine>,
+}
+
+#[allow(clippy::print_stdout)]
+impl Repl {
+    fn print_help(&self) {
+        println!("{}", ReplCommand::help())
+    }
+
+    pub(crate) async fn try_new(cmd: &AttachCommand) -> Result<Self> {
+        let mut rl = Editor::new().context(ReplCreationSnafu)?;
+
+        if !cmd.disable_helper {
+            rl.set_helper(Some(RustylineHelper::default()));
+
+            let history_file = history_file();
+            if let Err(e) = rl.load_history(&history_file) {
+                logging::debug!(
+                    "failed to load history file on {}, error: {e}",
+                    history_file.display()
+                );
+            }
+        }
+
+        let client = Client::with_urls([&cmd.grpc_addr]);
+        let database = Database::new(DEFAULT_CATALOG_NAME, DEFAULT_SCHEMA_NAME, client);
+
+        let query_engine = if let Some(meta_addr) = &cmd.meta_addr {
+            create_query_engine(meta_addr).await.map(Some)?
+        } else {
+            None
+        };
+
+        Ok(Self {
+            rl,
+            prompt: "> ".to_string(),
+            database,
+            query_engine,
+        })
+    }
+
+    /// Parse the next command
+    fn next_command(&mut self) -> Result<ReplCommand> {
+        match self.rl.readline(&self.prompt) {
+            Ok(ref line) => {
+                let request = line.trim();
+
+                self.rl.add_history_entry(request.to_string());
+
+                request.try_into()
+            }
+            Err(ReadlineError::Eof) | Err(ReadlineError::Interrupted) => Ok(ReplCommand::Exit),
+            // Some sort of real underlying error
+            Err(e) => Err(e).context(ReadlineSnafu),
+        }
+    }
+
+    /// Read Evaluate Print Loop (interactive command line) for GreptimeDB
+    ///
+    /// Inspired / based on repl.rs from InfluxDB IOX
+    pub(crate) async fn run(&mut self) -> Result<()> {
+        println!("Ready for commands. (Hint: try 'help')");
+
+        loop {
+            match self.next_command()? {
+                ReplCommand::Help => {
+                    self.print_help();
+                }
+                ReplCommand::UseDatabase { db_name } => {
+                    if self.execute_sql(format!("USE {db_name}")).await {
+                        println!("Using {db_name}");
+                        self.database.set_schema(&db_name);
+                        self.prompt = format!("[{db_name}] > ");
+                    }
+                }
+                ReplCommand::Sql { sql } => {
+                    self.execute_sql(sql).await;
+                }
+                ReplCommand::Exit => {
+                    return Ok(());
+                }
+            }
+        }
+    }
+
+    async fn execute_sql(&self, sql: String) -> bool {
+        self.do_execute_sql(sql)
+            .await
+            .map_err(|e| {
+                let status_code = e.status_code();
+                let root_cause = e.iter_chain().last().unwrap();
+                println!("Error: {}({status_code}), {root_cause}", status_code as u32)
+            })
+            .is_ok()
+    }
+
+    async fn do_execute_sql(&self, sql: String) -> Result<()> {
+        let start = Instant::now();
+
+        let output = if let Some(query_engine) = &self.query_engine {
+            let stmt = QueryLanguageParser::parse_sql(&sql)
+                .with_context(|_| ParseSqlSnafu { sql: sql.clone() })?;
+
+            let query_ctx = Arc::new(QueryContext::with(
+                self.database.catalog(),
+                self.database.schema(),
+            ));
+
+            let plan = query_engine
+                .planner()
+                .plan(stmt, query_ctx)
+                .await
+                .context(PlanStatementSnafu)?;
+
+            let LogicalPlan::DfPlan(plan) =
+                query_engine.optimize(&plan).context(PlanStatementSnafu)?;
+
+            let plan = DFLogicalSubstraitConvertor {}
+                .encode(plan)
+                .context(SubstraitEncodeLogicalPlanSnafu)?;
+
+            self.database.logical_plan(plan.to_vec()).await
+        } else {
+            self.database.sql(&sql).await
+        }
+        .context(RequestDatabaseSnafu { sql: &sql })?;
+
+        let either = match output {
+            Output::Stream(s) => {
+                let x = RecordBatches::try_collect(s)
+                    .await
+                    .context(CollectRecordBatchesSnafu)?;
+                Either::Left(x)
+            }
+            Output::RecordBatches(x) => Either::Left(x),
+            Output::AffectedRows(rows) => Either::Right(rows),
+        };
+
+        let end = Instant::now();
+
+        match either {
+            Either::Left(recordbatches) => {
+                let total_rows: usize = recordbatches.iter().map(|x| x.num_rows()).sum();
+                if total_rows > 0 {
+                    println!(
+                        "{}",
+                        recordbatches
+                            .pretty_print()
+                            .context(PrettyPrintRecordBatchesSnafu)?
+                    );
+                }
+                println!("Total Rows: {total_rows}")
+            }
+            Either::Right(rows) => println!("Affected Rows: {rows}"),
+        };
+
+        println!("Cost {} ms", (end - start).as_millis());
+        Ok(())
+    }
+}
+
+impl Drop for Repl {
+    fn drop(&mut self) {
+        if self.rl.helper().is_some() {
+            let history_file = history_file();
+            if let Err(e) = self.rl.save_history(&history_file) {
+                logging::debug!(
+                    "failed to save history file on {}, error: {e}",
+                    history_file.display()
+                );
+            }
+        }
+    }
+}
+
+/// Return the location of the history file (defaults to $HOME/".greptimedb_cli_history")
+fn history_file() -> PathBuf {
+    let mut buf = match std::env::var("HOME") {
+        Ok(home) => PathBuf::from(home),
+        Err(_) => PathBuf::new(),
+    };
+    buf.push(".greptimedb_cli_history");
+    buf
+}
+
+async fn create_query_engine(meta_addr: &str) -> Result<DatafusionQueryEngine> {
+    let mut meta_client = MetaClientBuilder::default().enable_store().build();
+    meta_client
+        .start([meta_addr])
+        .await
+        .context(StartMetaClientSnafu)?;
+    let meta_client = Arc::new(meta_client);
+
+    let backend = Arc::new(MetaKvBackend {
+        client: meta_client.clone(),
+    });
+
+    let table_routes = Arc::new(TableRoutes::new(meta_client));
+    let partition_manager = Arc::new(PartitionRuleManager::new(table_routes));
+
+    let datanode_clients = Arc::new(DatanodeClients::default());
+
+    let catalog_list = Arc::new(FrontendCatalogManager::new(
+        backend,
+        partition_manager,
+        datanode_clients,
+    ));
+    let state = Arc::new(QueryEngineState::new(catalog_list, Default::default()));
+
+    Ok(DatafusionQueryEngine::new(state))
+}
--- a/src/cmd/src/datanode.rs
+++ b/src/cmd/src/datanode.rs
@@ -12,16 +12,37 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

+use std::time::Duration;
+
 use clap::Parser;
 use common_telemetry::logging;
-use datanode::datanode::{Datanode, DatanodeOptions, FileConfig, ObjectStoreConfig};
-use meta_client::MetaClientOpts;
+use datanode::datanode::{
+    Datanode, DatanodeOptions, FileConfig, ObjectStoreConfig, ProcedureConfig,
+};
+use meta_client::MetaClientOptions;
 use servers::Mode;
 use snafu::ResultExt;

-use crate::error::{Error, MissingConfigSnafu, Result, StartDatanodeSnafu};
+use crate::error::{Error, MissingConfigSnafu, Result, ShutdownDatanodeSnafu, StartDatanodeSnafu};
 use crate::toml_loader;

+pub struct Instance {
+    datanode: Datanode,
+}
+
+impl Instance {
+    pub async fn run(&mut self) -> Result<()> {
+        self.datanode.start().await.context(StartDatanodeSnafu)
+    }
+
+    pub async fn stop(&self) -> Result<()> {
+        self.datanode
+            .shutdown()
+            .await
+            .context(ShutdownDatanodeSnafu)
+    }
+}
+
 #[derive(Parser)]
 pub struct Command {
    #[clap(subcommand)]
@@ -29,8 +50,8 @@ pub struct Command {
 }

 impl Command {
-    pub async fn run(self) -> Result<()> {
-        self.subcmd.run().await
+    pub async fn build(self) -> Result<Instance> {
+        self.subcmd.build().await
    }
 }

@@ -40,9 +61,9 @@ enum SubCommand {
 }

 impl SubCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        match self {
-            SubCommand::Start(cmd) => cmd.run().await,
+            SubCommand::Start(cmd) => cmd.build().await,
        }
    }
 }
@@ -65,22 +86,25 @@ struct StartCommand {
    data_dir: Option<String>,
    #[clap(long)]
    wal_dir: Option<String>,
+    #[clap(long)]
+    procedure_dir: Option<String>,
+    #[clap(long)]
+    http_addr: Option<String>,
+    #[clap(long)]
+    http_timeout: Option<u64>,
 }

 impl StartCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        logging::info!("Datanode start command: {:#?}", self);

        let opts: DatanodeOptions = self.try_into()?;

        logging::info!("Datanode options: {:#?}", opts);

-        Datanode::new(opts)
-            .await
-            .context(StartDatanodeSnafu)?
-            .start()
-            .await
-            .context(StartDatanodeSnafu)
+        let datanode = Datanode::new(opts).await.context(StartDatanodeSnafu)?;
+
+        Ok(Instance { datanode })
    }
 }

@@ -110,8 +134,8 @@ impl TryFrom<StartCommand> for DatanodeOptions {
        }

        if let Some(meta_addr) = cmd.metasrv_addr {
-            opts.meta_client_opts
-                .get_or_insert_with(MetaClientOpts::default)
+            opts.meta_client_options
+                .get_or_insert_with(MetaClientOptions::default)
                .metasrv_addrs = meta_addr
                .split(',')
                .map(&str::trim)
@@ -128,12 +152,25 @@ impl TryFrom<StartCommand> for DatanodeOptions {
        }

        if let Some(data_dir) = cmd.data_dir {
-            opts.storage = ObjectStoreConfig::File(FileConfig { data_dir });
+            opts.storage.store = ObjectStoreConfig::File(FileConfig { data_dir });
        }

        if let Some(wal_dir) = cmd.wal_dir {
            opts.wal.dir = wal_dir;
        }
+        if let Some(procedure_dir) = cmd.procedure_dir {
+            opts.procedure = Some(ProcedureConfig::from_file_path(procedure_dir));
+        }
+        if let Some(http_addr) = cmd.http_addr {
+            opts.http_opts.addr = http_addr
+        }
+        if let Some(http_timeout) = cmd.http_timeout {
+            opts.http_opts.timeout = Duration::from_secs(http_timeout)
+        }
+
+        // Disable dashboard in datanode.
+        opts.http_opts.disable_dashboard = true;
+
        Ok(opts)
    }
 }
@@ -141,42 +178,89 @@ impl TryFrom<StartCommand> for DatanodeOptions {
 #[cfg(test)]
 mod tests {
    use std::assert_matches::assert_matches;
+    use std::io::Write;
    use std::time::Duration;

-    use datanode::datanode::{CompactionConfig, ObjectStoreConfig};
+    use common_base::readable_size::ReadableSize;
+    use common_test_util::temp_dir::create_named_temp_file;
+    use datanode::datanode::{CompactionConfig, ObjectStoreConfig, RegionManifestConfig};
    use servers::Mode;

    use super::*;

    #[test]
    fn test_read_from_config_file() {
+        let mut file = create_named_temp_file();
+        let toml_str = r#"
+            mode = "distributed"
+            enable_memory_catalog = false
+            node_id = 42
+            rpc_addr = "127.0.0.1:3001"
+            rpc_hostname = "127.0.0.1"
+            rpc_runtime_size = 8
+            mysql_addr = "127.0.0.1:4406"
+            mysql_runtime_size = 2
+
+            [meta_client_options]
+            metasrv_addrs = ["127.0.0.1:3002"]
+            timeout_millis = 3000
+            connect_timeout_millis = 5000
+            tcp_nodelay = true
+
+            [wal]
+            dir = "/tmp/greptimedb/wal"
+            file_size = "1GB"
+            purge_threshold = "50GB"
+            purge_interval = "10m"
+            read_batch_size = 128
+            sync_write = false
+
+            [storage]
+            type = "File"
+            data_dir = "/tmp/greptimedb/data/"
+
+            [storage.compaction]
+            max_inflight_tasks = 3
+            max_files_in_level0 = 7
+            max_purge_tasks = 32
+
+            [storage.manifest]
+            checkpoint_margin = 9
+            gc_duration = '7s'
+            checkpoint_on_startup = true
+        "#;
+        write!(file, "{}", toml_str).unwrap();
+
        let cmd = StartCommand {
-            config_file: Some(format!(
-                "{}/../../config/datanode.example.toml",
-                std::env::current_dir().unwrap().as_path().to_str().unwrap()
-            )),
+            config_file: Some(file.path().to_str().unwrap().to_string()),
            ..Default::default()
        };
        let options: DatanodeOptions = cmd.try_into().unwrap();
        assert_eq!("127.0.0.1:3001".to_string(), options.rpc_addr);
-        assert_eq!("/tmp/greptimedb/wal".to_string(), options.wal.dir);
        assert_eq!("127.0.0.1:4406".to_string(), options.mysql_addr);
-        assert_eq!(4, options.mysql_runtime_size);
-        let MetaClientOpts {
+        assert_eq!(2, options.mysql_runtime_size);
+        assert_eq!(Some(42), options.node_id);
+
+        assert_eq!(Duration::from_secs(600), options.wal.purge_interval);
+        assert_eq!(1024 * 1024 * 1024, options.wal.file_size.0);
+        assert_eq!(1024 * 1024 * 1024 * 50, options.wal.purge_threshold.0);
+        assert!(!options.wal.sync_write);
+
+        let MetaClientOptions {
            metasrv_addrs: metasrv_addr,
            timeout_millis,
            connect_timeout_millis,
            tcp_nodelay,
-        } = options.meta_client_opts.unwrap();
+        } = options.meta_client_options.unwrap();

        assert_eq!(vec!["127.0.0.1:3002".to_string()], metasrv_addr);
        assert_eq!(5000, connect_timeout_millis);
        assert_eq!(3000, timeout_millis);
-        assert!(!tcp_nodelay);
+        assert!(tcp_nodelay);

-        match options.storage {
-            ObjectStoreConfig::File(FileConfig { data_dir }) => {
-                assert_eq!("/tmp/greptimedb/data/".to_string(), data_dir)
+        match &options.storage.store {
+            ObjectStoreConfig::File(FileConfig { data_dir, .. }) => {
+                assert_eq!("/tmp/greptimedb/data/", data_dir)
            }
            ObjectStoreConfig::S3 { .. } => unreachable!(),
            ObjectStoreConfig::Oss { .. } => unreachable!(),
@@ -184,11 +268,20 @@ mod tests {

        assert_eq!(
            CompactionConfig {
-                max_inflight_tasks: 4,
-                max_files_in_level0: 16,
+                max_inflight_tasks: 3,
+                max_files_in_level0: 7,
                max_purge_tasks: 32,
+                sst_write_buffer_size: ReadableSize::mb(8),
            },
-            options.compaction
+            options.storage.compaction,
+        );
+        assert_eq!(
+            RegionManifestConfig {
+                checkpoint_margin: Some(9),
+                gc_duration: Some(Duration::from_secs(7)),
+                checkpoint_on_startup: true,
+            },
+            options.storage.manifest,
        );
    }

@@ -223,32 +316,4 @@ mod tests {
        })
        .unwrap();
    }
-
-    #[test]
-    fn test_merge_config() {
-        let dn_opts = DatanodeOptions::try_from(StartCommand {
-            config_file: Some(format!(
-                "{}/../../config/datanode.example.toml",
-                std::env::current_dir().unwrap().as_path().to_str().unwrap()
-            )),
-            ..Default::default()
-        })
-        .unwrap();
-        assert_eq!("/tmp/greptimedb/wal", dn_opts.wal.dir);
-        assert_eq!(Duration::from_secs(600), dn_opts.wal.purge_interval);
-        assert_eq!(1024 * 1024 * 1024, dn_opts.wal.file_size.0);
-        assert_eq!(1024 * 1024 * 1024 * 50, dn_opts.wal.purge_threshold.0);
-        assert!(!dn_opts.wal.sync_write);
-        assert_eq!(Some(42), dn_opts.node_id);
-        let MetaClientOpts {
-            metasrv_addrs: metasrv_addr,
-            timeout_millis,
-            connect_timeout_millis,
-            tcp_nodelay,
-        } = dn_opts.meta_client_opts.unwrap();
-        assert_eq!(vec!["127.0.0.1:3002".to_string()], metasrv_addr);
-        assert_eq!(3000, timeout_millis);
-        assert_eq!(5000, connect_timeout_millis);
-        assert!(!tcp_nodelay);
-    }
 }
--- a/src/cmd/src/error.rs
+++ b/src/cmd/src/error.rs
@@ -15,6 +15,8 @@
 use std::any::Any;

 use common_error::prelude::*;
+use rustyline::error::ReadlineError;
+use snafu::Location;

 #[derive(Debug, Snafu)]
 #[snafu(visibility(pub))]
@@ -25,36 +27,60 @@ pub enum Error {
        source: datanode::error::Error,
    },

+    #[snafu(display("Failed to shutdown datanode, source: {}", source))]
+    ShutdownDatanode {
+        #[snafu(backtrace)]
+        source: datanode::error::Error,
+    },
+
    #[snafu(display("Failed to start frontend, source: {}", source))]
    StartFrontend {
        #[snafu(backtrace)]
        source: frontend::error::Error,
    },

+    #[snafu(display("Failed to shutdown frontend, source: {}", source))]
+    ShutdownFrontend {
+        #[snafu(backtrace)]
+        source: frontend::error::Error,
+    },
+
+    #[snafu(display("Failed to build meta server, source: {}", source))]
+    BuildMetaServer {
+        #[snafu(backtrace)]
+        source: meta_srv::error::Error,
+    },
+
    #[snafu(display("Failed to start meta server, source: {}", source))]
    StartMetaServer {
        #[snafu(backtrace)]
        source: meta_srv::error::Error,
    },

+    #[snafu(display("Failed to shutdown meta server, source: {}", source))]
+    ShutdownMetaServer {
+        #[snafu(backtrace)]
+        source: meta_srv::error::Error,
+    },
+
    #[snafu(display("Failed to read config file: {}, source: {}", path, source))]
    ReadConfig {
        path: String,
        source: std::io::Error,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Failed to parse config, source: {}", source))]
    ParseConfig {
        source: toml::de::Error,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Missing config, msg: {}", msg))]
-    MissingConfig { msg: String, backtrace: Backtrace },
+    MissingConfig { msg: String, location: Location },

    #[snafu(display("Illegal config: {}", msg))]
-    IllegalConfig { msg: String, backtrace: Backtrace },
+    IllegalConfig { msg: String, location: Location },

    #[snafu(display("Illegal auth config: {}", source))]
    IllegalAuthConfig {
@@ -68,6 +94,65 @@ pub enum Error {
        #[snafu(backtrace)]
        source: meta_srv::error::Error,
    },
+
+    #[snafu(display("Invalid REPL command: {reason}"))]
+    InvalidReplCommand { reason: String },
+
+    #[snafu(display("Cannot create REPL: {}", source))]
+    ReplCreation {
+        source: ReadlineError,
+        location: Location,
+    },
+
+    #[snafu(display("Error reading command: {}", source))]
+    Readline {
+        source: ReadlineError,
+        location: Location,
+    },
+
+    #[snafu(display("Failed to request database, sql: {sql}, source: {source}"))]
+    RequestDatabase {
+        sql: String,
+        #[snafu(backtrace)]
+        source: client::Error,
+    },
+
+    #[snafu(display("Failed to collect RecordBatches, source: {source}"))]
+    CollectRecordBatches {
+        #[snafu(backtrace)]
+        source: common_recordbatch::error::Error,
+    },
+
+    #[snafu(display("Failed to pretty print Recordbatches, source: {source}"))]
+    PrettyPrintRecordBatches {
+        #[snafu(backtrace)]
+        source: common_recordbatch::error::Error,
+    },
+
+    #[snafu(display("Failed to start Meta client, source: {}", source))]
+    StartMetaClient {
+        #[snafu(backtrace)]
+        source: meta_client::error::Error,
+    },
+
+    #[snafu(display("Failed to parse SQL: {}, source: {}", sql, source))]
+    ParseSql {
+        sql: String,
+        #[snafu(backtrace)]
+        source: query::error::Error,
+    },
+
+    #[snafu(display("Failed to plan statement, source: {}", source))]
+    PlanStatement {
+        #[snafu(backtrace)]
+        source: query::error::Error,
+    },
+
+    #[snafu(display("Failed to encode logical plan in substrait, source: {}", source))]
+    SubstraitEncodeLogicalPlan {
+        #[snafu(backtrace)]
+        source: substrait::error::Error,
+    },
 }

 pub type Result<T> = std::result::Result<T, Error>;
@@ -77,88 +162,33 @@ impl ErrorExt for Error {
        match self {
            Error::StartDatanode { source } => source.status_code(),
            Error::StartFrontend { source } => source.status_code(),
+            Error::ShutdownDatanode { source } => source.status_code(),
+            Error::ShutdownFrontend { source } => source.status_code(),
            Error::StartMetaServer { source } => source.status_code(),
+            Error::ShutdownMetaServer { source } => source.status_code(),
+            Error::BuildMetaServer { source } => source.status_code(),
            Error::UnsupportedSelectorType { source, .. } => source.status_code(),
            Error::ReadConfig { .. } | Error::ParseConfig { .. } | Error::MissingConfig { .. } => {
                StatusCode::InvalidArguments
            }
-            Error::IllegalConfig { .. } => StatusCode::InvalidArguments,
+            Error::IllegalConfig { .. } | Error::InvalidReplCommand { .. } => {
+                StatusCode::InvalidArguments
+            }
            Error::IllegalAuthConfig { .. } => StatusCode::InvalidArguments,
+            Error::ReplCreation { .. } | Error::Readline { .. } => StatusCode::Internal,
+            Error::RequestDatabase { source, .. } => source.status_code(),
+            Error::CollectRecordBatches { source } | Error::PrettyPrintRecordBatches { source } => {
+                source.status_code()
+            }
+            Error::StartMetaClient { source } => source.status_code(),
+            Error::ParseSql { source, .. } | Error::PlanStatement { source } => {
+                source.status_code()
+            }
+            Error::SubstraitEncodeLogicalPlan { source } => source.status_code(),
        }
    }

-    fn backtrace_opt(&self) -> Option<&Backtrace> {
-        ErrorCompat::backtrace(self)
-    }
-
    fn as_any(&self) -> &dyn Any {
        self
    }
 }
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    type StdResult<E> = std::result::Result<(), E>;
-
-    #[test]
-    fn test_start_node_error() {
-        fn throw_datanode_error() -> StdResult<datanode::error::Error> {
-            datanode::error::MissingNodeIdSnafu {}.fail()
-        }
-
-        let e = throw_datanode_error()
-            .context(StartDatanodeSnafu)
-            .err()
-            .unwrap();
-
-        assert!(e.backtrace_opt().is_some());
-        assert_eq!(e.status_code(), StatusCode::InvalidArguments);
-    }
-
-    #[test]
-    fn test_start_frontend_error() {
-        fn throw_frontend_error() -> StdResult<frontend::error::Error> {
-            frontend::error::InvalidSqlSnafu { err_msg: "failed" }.fail()
-        }
-
-        let e = throw_frontend_error()
-            .context(StartFrontendSnafu)
-            .err()
-            .unwrap();
-
-        assert!(e.backtrace_opt().is_some());
-        assert_eq!(e.status_code(), StatusCode::InvalidArguments);
-    }
-
-    #[test]
-    fn test_start_metasrv_error() {
-        fn throw_metasrv_error() -> StdResult<meta_srv::error::Error> {
-            meta_srv::error::StreamNoneSnafu {}.fail()
-        }
-
-        let e = throw_metasrv_error()
-            .context(StartMetaServerSnafu)
-            .err()
-            .unwrap();
-
-        assert!(e.backtrace_opt().is_some());
-        assert_eq!(e.status_code(), StatusCode::Internal);
-    }
-
-    #[test]
-    fn test_read_config_error() {
-        fn throw_read_config_error() -> StdResult<std::io::Error> {
-            Err(std::io::ErrorKind::NotFound.into())
-        }
-
-        let e = throw_read_config_error()
-            .context(ReadConfigSnafu { path: "test" })
-            .err()
-            .unwrap();
-
-        assert!(e.backtrace_opt().is_some());
-        assert_eq!(e.status_code(), StatusCode::InvalidArguments);
-    }
-}
--- a/src/cmd/src/frontend.rs
+++ b/src/cmd/src/frontend.rs
@@ -16,14 +16,15 @@ use std::sync::Arc;

 use clap::Parser;
 use common_base::Plugins;
-use frontend::frontend::{Frontend, FrontendOptions};
+use frontend::frontend::FrontendOptions;
 use frontend::grpc::GrpcOptions;
 use frontend::influxdb::InfluxdbOptions;
-use frontend::instance::Instance;
+use frontend::instance::{FrontendInstance, Instance as FeInstance};
 use frontend::mysql::MysqlOptions;
 use frontend::opentsdb::OpentsdbOptions;
 use frontend::postgres::PostgresOptions;
-use meta_client::MetaClientOpts;
+use frontend::prom::PromOptions;
+use meta_client::MetaClientOptions;
 use servers::auth::UserProviderRef;
 use servers::http::HttpOptions;
 use servers::tls::{TlsMode, TlsOption};
@@ -33,6 +34,26 @@ use snafu::ResultExt;
 use crate::error::{self, IllegalAuthConfigSnafu, Result};
 use crate::toml_loader;

+pub struct Instance {
+    frontend: FeInstance,
+}
+
+impl Instance {
+    pub async fn run(&mut self) -> Result<()> {
+        self.frontend
+            .start()
+            .await
+            .context(error::StartFrontendSnafu)
+    }
+
+    pub async fn stop(&self) -> Result<()> {
+        self.frontend
+            .shutdown()
+            .await
+            .context(error::ShutdownFrontendSnafu)
+    }
+}
+
 #[derive(Parser)]
 pub struct Command {
    #[clap(subcommand)]
@@ -40,8 +61,8 @@ pub struct Command {
 }

 impl Command {
-    pub async fn run(self) -> Result<()> {
-        self.subcmd.run().await
+    pub async fn build(self) -> Result<Instance> {
+        self.subcmd.build().await
    }
 }

@@ -51,9 +72,9 @@ enum SubCommand {
 }

 impl SubCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        match self {
-            SubCommand::Start(cmd) => cmd.run().await,
+            SubCommand::Start(cmd) => cmd.build().await,
        }
    }
 }
@@ -67,6 +88,8 @@ pub struct StartCommand {
    #[clap(long)]
    mysql_addr: Option<String>,
    #[clap(long)]
+    prom_addr: Option<String>,
+    #[clap(long)]
    postgres_addr: Option<String>,
    #[clap(long)]
    opentsdb_addr: Option<String>,
@@ -84,19 +107,25 @@ pub struct StartCommand {
    tls_key_path: Option<String>,
    #[clap(long)]
    user_provider: Option<String>,
+    #[clap(long)]
+    disable_dashboard: bool,
 }

 impl StartCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        let plugins = Arc::new(load_frontend_plugins(&self.user_provider)?);
        let opts: FrontendOptions = self.try_into()?;

-        let instance = Instance::try_new_distributed(&opts, plugins.clone())
+        let mut instance = FeInstance::try_new_distributed(&opts, plugins.clone())
            .await
            .context(error::StartFrontendSnafu)?;

-        let mut frontend = Frontend::new(opts, instance, plugins);
-        frontend.start().await.context(error::StartFrontendSnafu)
+        instance
+            .build_servers(&opts, plugins)
+            .await
+            .context(error::StartFrontendSnafu)?;
+
+        Ok(Instance { frontend: instance })
    }
 }

@@ -122,18 +151,24 @@ impl TryFrom<StartCommand> for FrontendOptions {

        let tls_option = TlsOption::new(cmd.tls_mode, cmd.tls_cert_path, cmd.tls_key_path);

+        let mut http_options = HttpOptions {
+            disable_dashboard: cmd.disable_dashboard,
+            ..Default::default()
+        };
+
        if let Some(addr) = cmd.http_addr {
-            opts.http_options = Some(HttpOptions {
-                addr,
-                ..Default::default()
-            });
+            http_options.addr = addr;
        }
+
+        opts.http_options = Some(http_options);
+
        if let Some(addr) = cmd.grpc_addr {
            opts.grpc_options = Some(GrpcOptions {
                addr,
                ..Default::default()
            });
        }
+
        if let Some(addr) = cmd.mysql_addr {
            opts.mysql_options = Some(MysqlOptions {
                addr,
@@ -141,6 +176,9 @@ impl TryFrom<StartCommand> for FrontendOptions {
                ..Default::default()
            });
        }
+        if let Some(addr) = cmd.prom_addr {
+            opts.prom_options = Some(PromOptions { addr });
+        }
        if let Some(addr) = cmd.postgres_addr {
            opts.postgres_options = Some(PostgresOptions {
                addr,
@@ -158,8 +196,8 @@ impl TryFrom<StartCommand> for FrontendOptions {
            opts.influxdb_options = Some(InfluxdbOptions { enable });
        }
        if let Some(metasrv_addr) = cmd.metasrv_addr {
-            opts.meta_client_opts
-                .get_or_insert_with(MetaClientOpts::default)
+            opts.meta_client_options
+                .get_or_insert_with(MetaClientOptions::default)
                .metasrv_addrs = metasrv_addr
                .split(',')
                .map(&str::trim)
@@ -173,8 +211,10 @@ impl TryFrom<StartCommand> for FrontendOptions {

 #[cfg(test)]
 mod tests {
+    use std::io::Write;
    use std::time::Duration;

+    use common_test_util::temp_dir::create_named_temp_file;
    use servers::auth::{Identity, Password, UserProviderRef};

    use super::*;
@@ -184,6 +224,7 @@ mod tests {
        let command = StartCommand {
            http_addr: Some("127.0.0.1:1234".to_string()),
            grpc_addr: None,
+            prom_addr: Some("127.0.0.1:4444".to_string()),
            mysql_addr: Some("127.0.0.1:5678".to_string()),
            postgres_addr: Some("127.0.0.1:5432".to_string()),
            opentsdb_addr: Some("127.0.0.1:4321".to_string()),
@@ -194,6 +235,7 @@ mod tests {
            tls_cert_path: None,
            tls_key_path: None,
            user_provider: None,
+            disable_dashboard: false,
        };

        let opts: FrontendOptions = command.try_into().unwrap();
@@ -207,6 +249,7 @@ mod tests {
            opts.opentsdb_options.as_ref().unwrap().addr,
            "127.0.0.1:4321"
        );
+        assert_eq!(opts.prom_options.as_ref().unwrap().addr, "127.0.0.1:4444");

        let default_opts = FrontendOptions::default();
        assert_eq!(
@@ -231,22 +274,31 @@ mod tests {

    #[test]
    fn test_read_from_config_file() {
+        let mut file = create_named_temp_file();
+        let toml_str = r#"
+            mode = "distributed"
+
+            [http_options]
+            addr = "127.0.0.1:4000"
+            timeout = "30s"
+        "#;
+        write!(file, "{}", toml_str).unwrap();
+
        let command = StartCommand {
            http_addr: None,
            grpc_addr: None,
            mysql_addr: None,
+            prom_addr: None,
            postgres_addr: None,
            opentsdb_addr: None,
            influxdb_enable: None,
-            config_file: Some(format!(
-                "{}/../../config/frontend.example.toml",
-                std::env::current_dir().unwrap().as_path().to_str().unwrap()
-            )),
+            config_file: Some(file.path().to_str().unwrap().to_string()),
            metasrv_addr: None,
            tls_mode: None,
            tls_cert_path: None,
            tls_key_path: None,
            user_provider: None,
+            disable_dashboard: false,
        };

        let fe_opts = FrontendOptions::try_from(command).unwrap();
@@ -267,6 +319,7 @@ mod tests {
            http_addr: None,
            grpc_addr: None,
            mysql_addr: None,
+            prom_addr: None,
            postgres_addr: None,
            opentsdb_addr: None,
            influxdb_enable: None,
@@ -276,6 +329,7 @@ mod tests {
            tls_cert_path: None,
            tls_key_path: None,
            user_provider: Some("static_user_provider:cmd:test=test".to_string()),
+            disable_dashboard: false,
        };

        let plugins = load_frontend_plugins(&command.user_provider);
--- a/src/cmd/src/lib.rs
+++ b/src/cmd/src/lib.rs
@@ -14,6 +14,7 @@

 #![feature(assert_matches)]

+pub mod cli;
 pub mod datanode;
 pub mod error;
 pub mod frontend;
--- a/src/cmd/src/metasrv.rs
+++ b/src/cmd/src/metasrv.rs
@@ -12,15 +12,37 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

+use std::time::Duration;
+
 use clap::Parser;
 use common_telemetry::{info, logging, warn};
-use meta_srv::bootstrap;
+use meta_srv::bootstrap::MetaSrvInstance;
 use meta_srv::metasrv::MetaSrvOptions;
 use snafu::ResultExt;

 use crate::error::{Error, Result};
 use crate::{error, toml_loader};

+pub struct Instance {
+    instance: MetaSrvInstance,
+}
+
+impl Instance {
+    pub async fn run(&mut self) -> Result<()> {
+        self.instance
+            .start()
+            .await
+            .context(error::StartMetaServerSnafu)
+    }
+
+    pub async fn stop(&self) -> Result<()> {
+        self.instance
+            .shutdown()
+            .await
+            .context(error::ShutdownMetaServerSnafu)
+    }
+}
+
 #[derive(Parser)]
 pub struct Command {
    #[clap(subcommand)]
@@ -28,8 +50,8 @@ pub struct Command {
 }

 impl Command {
-    pub async fn run(self) -> Result<()> {
-        self.subcmd.run().await
+    pub async fn build(self) -> Result<Instance> {
+        self.subcmd.build().await
    }
 }

@@ -39,9 +61,9 @@ enum SubCommand {
 }

 impl SubCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        match self {
-            SubCommand::Start(cmd) => cmd.run().await,
+            SubCommand::Start(cmd) => cmd.build().await,
        }
    }
 }
@@ -60,19 +82,24 @@ struct StartCommand {
    selector: Option<String>,
    #[clap(long)]
    use_memory_store: bool,
+    #[clap(long)]
+    http_addr: Option<String>,
+    #[clap(long)]
+    http_timeout: Option<u64>,
 }

 impl StartCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        logging::info!("MetaSrv start command: {:#?}", self);

        let opts: MetaSrvOptions = self.try_into()?;

        logging::info!("MetaSrv options: {:#?}", opts);
-
-        bootstrap::bootstrap_meta_srv(opts)
+        let instance = MetaSrvInstance::new(opts)
            .await
-            .context(error::StartMetaServerSnafu)
+            .context(error::BuildMetaServerSnafu)?;
+
+        Ok(Instance { instance })
    }
 }

@@ -107,12 +134,25 @@ impl TryFrom<StartCommand> for MetaSrvOptions {
            opts.use_memory_store = true;
        }

+        if let Some(http_addr) = cmd.http_addr {
+            opts.http_opts.addr = http_addr;
+        }
+        if let Some(http_timeout) = cmd.http_timeout {
+            opts.http_opts.timeout = Duration::from_secs(http_timeout);
+        }
+
+        // Disable dashboard in metasrv.
+        opts.http_opts.disable_dashboard = true;
+
        Ok(opts)
    }
 }

 #[cfg(test)]
 mod tests {
+    use std::io::Write;
+
+    use common_test_util::temp_dir::create_named_temp_file;
    use meta_srv::selector::SelectorType;

    use super::*;
@@ -126,6 +166,8 @@ mod tests {
            config_file: None,
            selector: Some("LoadBased".to_string()),
            use_memory_store: false,
+            http_addr: None,
+            http_timeout: None,
        };
        let options: MetaSrvOptions = cmd.try_into().unwrap();
        assert_eq!("127.0.0.1:3002".to_string(), options.bind_addr);
@@ -136,16 +178,26 @@ mod tests {

    #[test]
    fn test_read_from_config_file() {
+        let mut file = create_named_temp_file();
+        let toml_str = r#"
+            bind_addr = "127.0.0.1:3002"
+            server_addr = "127.0.0.1:3002"
+            store_addr = "127.0.0.1:2379"
+            datanode_lease_secs = 15
+            selector = "LeaseBased"
+            use_memory_store = false
+        "#;
+        write!(file, "{}", toml_str).unwrap();
+
        let cmd = StartCommand {
            bind_addr: None,
            server_addr: None,
            store_addr: None,
            selector: None,
-            config_file: Some(format!(
-                "{}/../../config/metasrv.example.toml",
-                std::env::current_dir().unwrap().as_path().to_str().unwrap()
-            )),
+            config_file: Some(file.path().to_str().unwrap().to_string()),
            use_memory_store: false,
+            http_addr: None,
+            http_timeout: None,
        };
        let options: MetaSrvOptions = cmd.try_into().unwrap();
        assert_eq!("127.0.0.1:3002".to_string(), options.bind_addr);
--- a/src/cmd/src/standalone.rs
+++ b/src/cmd/src/standalone.rs
@@ -17,24 +17,27 @@ use std::sync::Arc;
 use clap::Parser;
 use common_base::Plugins;
 use common_telemetry::info;
-use datanode::datanode::{Datanode, DatanodeOptions, ObjectStoreConfig, WalConfig};
+use datanode::datanode::{Datanode, DatanodeOptions, ProcedureConfig, StorageConfig, WalConfig};
 use datanode::instance::InstanceRef;
-use frontend::frontend::{Frontend, FrontendOptions};
+use frontend::frontend::FrontendOptions;
 use frontend::grpc::GrpcOptions;
 use frontend::influxdb::InfluxdbOptions;
-use frontend::instance::Instance as FeInstance;
+use frontend::instance::{FrontendInstance, Instance as FeInstance};
 use frontend::mysql::MysqlOptions;
 use frontend::opentsdb::OpentsdbOptions;
 use frontend::postgres::PostgresOptions;
+use frontend::prom::PromOptions;
 use frontend::prometheus::PrometheusOptions;
-use frontend::promql::PromqlOptions;
 use serde::{Deserialize, Serialize};
 use servers::http::HttpOptions;
 use servers::tls::{TlsMode, TlsOption};
 use servers::Mode;
 use snafu::ResultExt;

-use crate::error::{Error, IllegalConfigSnafu, Result, StartDatanodeSnafu, StartFrontendSnafu};
+use crate::error::{
+    Error, IllegalConfigSnafu, Result, ShutdownDatanodeSnafu, ShutdownFrontendSnafu,
+    StartDatanodeSnafu, StartFrontendSnafu,
+};
 use crate::frontend::load_frontend_plugins;
 use crate::toml_loader;

@@ -45,8 +48,8 @@ pub struct Command {
 }

 impl Command {
-    pub async fn run(self) -> Result<()> {
-        self.subcmd.run().await
+    pub async fn build(self) -> Result<Instance> {
+        self.subcmd.build().await
    }
 }

@@ -56,9 +59,9 @@ enum SubCommand {
 }

 impl SubCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        match self {
-            SubCommand::Start(cmd) => cmd.run().await,
+            SubCommand::Start(cmd) => cmd.build().await,
        }
    }
 }
@@ -66,6 +69,8 @@ impl SubCommand {
 #[derive(Clone, Debug, Serialize, Deserialize)]
 #[serde(default)]
 pub struct StandaloneOptions {
+    pub mode: Mode,
+    pub enable_memory_catalog: bool,
    pub http_options: Option<HttpOptions>,
    pub grpc_options: Option<GrpcOptions>,
    pub mysql_options: Option<MysqlOptions>,
@@ -73,16 +78,17 @@ pub struct StandaloneOptions {
    pub opentsdb_options: Option<OpentsdbOptions>,
    pub influxdb_options: Option<InfluxdbOptions>,
    pub prometheus_options: Option<PrometheusOptions>,
-    pub promql_options: Option<PromqlOptions>,
-    pub mode: Mode,
+    pub prom_options: Option<PromOptions>,
    pub wal: WalConfig,
-    pub storage: ObjectStoreConfig,
-    pub enable_memory_catalog: bool,
+    pub storage: StorageConfig,
+    pub procedure: Option<ProcedureConfig>,
 }

 impl Default for StandaloneOptions {
    fn default() -> Self {
        Self {
+            mode: Mode::Standalone,
+            enable_memory_catalog: false,
            http_options: Some(HttpOptions::default()),
            grpc_options: Some(GrpcOptions::default()),
            mysql_options: Some(MysqlOptions::default()),
@@ -90,11 +96,10 @@ impl Default for StandaloneOptions {
            opentsdb_options: Some(OpentsdbOptions::default()),
            influxdb_options: Some(InfluxdbOptions::default()),
            prometheus_options: Some(PrometheusOptions::default()),
-            promql_options: Some(PromqlOptions::default()),
-            mode: Mode::Standalone,
+            prom_options: Some(PromOptions::default()),
            wal: WalConfig::default(),
-            storage: ObjectStoreConfig::default(),
-            enable_memory_catalog: false,
+            storage: StorageConfig::default(),
+            procedure: None,
        }
    }
 }
@@ -102,6 +107,7 @@ impl Default for StandaloneOptions {
 impl StandaloneOptions {
    fn frontend_options(self) -> FrontendOptions {
        FrontendOptions {
+            mode: self.mode,
            http_options: self.http_options,
            grpc_options: self.grpc_options,
            mysql_options: self.mysql_options,
@@ -109,22 +115,56 @@ impl StandaloneOptions {
            opentsdb_options: self.opentsdb_options,
            influxdb_options: self.influxdb_options,
            prometheus_options: self.prometheus_options,
-            promql_options: self.promql_options,
-            mode: self.mode,
-            meta_client_opts: None,
+            prom_options: self.prom_options,
+            meta_client_options: None,
        }
    }

    fn datanode_options(self) -> DatanodeOptions {
        DatanodeOptions {
+            enable_memory_catalog: self.enable_memory_catalog,
            wal: self.wal,
            storage: self.storage,
-            enable_memory_catalog: self.enable_memory_catalog,
+            procedure: self.procedure,
            ..Default::default()
        }
    }
 }

+pub struct Instance {
+    datanode: Datanode,
+    frontend: FeInstance,
+}
+
+impl Instance {
+    pub async fn run(&mut self) -> Result<()> {
+        // Start datanode instance before starting services, to avoid requests come in before internal components are started.
+        self.datanode
+            .start_instance()
+            .await
+            .context(StartDatanodeSnafu)?;
+        info!("Datanode instance started");
+
+        self.frontend.start().await.context(StartFrontendSnafu)?;
+        Ok(())
+    }
+
+    pub async fn stop(&self) -> Result<()> {
+        self.frontend
+            .shutdown()
+            .await
+            .context(ShutdownFrontendSnafu)?;
+
+        self.datanode
+            .shutdown_instance()
+            .await
+            .context(ShutdownDatanodeSnafu)?;
+        info!("Datanode instance stopped.");
+
+        Ok(())
+    }
+}
+
 #[derive(Debug, Parser)]
 struct StartCommand {
    #[clap(long)]
@@ -134,6 +174,8 @@ struct StartCommand {
    #[clap(long)]
    mysql_addr: Option<String>,
    #[clap(long)]
+    prom_addr: Option<String>,
+    #[clap(long)]
    postgres_addr: Option<String>,
    #[clap(long)]
    opentsdb_addr: Option<String>,
@@ -154,7 +196,7 @@ struct StartCommand {
 }

 impl StartCommand {
-    async fn run(self) -> Result<()> {
+    async fn build(self) -> Result<Instance> {
        let enable_memory_catalog = self.enable_memory_catalog;
        let config_file = self.config_file.clone();
        let plugins = Arc::new(load_frontend_plugins(&self.user_provider)?);
@@ -174,33 +216,31 @@ impl StartCommand {
            fe_opts, dn_opts
        );

-        let mut datanode = Datanode::new(dn_opts.clone())
+        let datanode = Datanode::new(dn_opts.clone())
            .await
            .context(StartDatanodeSnafu)?;
-        let mut frontend = build_frontend(fe_opts, plugins, datanode.get_instance()).await?;

-        // Start datanode instance before starting services, to avoid requests come in before internal components are started.
-        datanode
-            .start_instance()
+        let mut frontend = build_frontend(plugins.clone(), datanode.get_instance()).await?;
+
+        frontend
+            .build_servers(&fe_opts, plugins)
            .await
-            .context(StartDatanodeSnafu)?;
-        info!("Datanode instance started");
+            .context(StartFrontendSnafu)?;

-        frontend.start().await.context(StartFrontendSnafu)?;
-        Ok(())
+        Ok(Instance { datanode, frontend })
    }
 }

 /// Build frontend instance in standalone mode
 async fn build_frontend(
-    fe_opts: FrontendOptions,
    plugins: Arc<Plugins>,
    datanode_instance: InstanceRef,
-) -> Result<Frontend<FeInstance>> {
-    let mut frontend_instance = FeInstance::new_standalone(datanode_instance.clone());
-    frontend_instance.set_script_handler(datanode_instance);
+) -> Result<FeInstance> {
+    let mut frontend_instance = FeInstance::try_new_standalone(datanode_instance.clone())
+        .await
+        .context(StartFrontendSnafu)?;
    frontend_instance.set_plugins(plugins.clone());
-    Ok(Frontend::new(fe_opts, frontend_instance, plugins))
+    Ok(frontend_instance)
 }

 impl TryFrom<StartCommand> for FrontendOptions {
@@ -246,6 +286,11 @@ impl TryFrom<StartCommand> for FrontendOptions {
                ..Default::default()
            })
        }
+
+        if let Some(addr) = cmd.prom_addr {
+            opts.prom_options = Some(PromOptions { addr })
+        }
+
        if let Some(addr) = cmd.postgres_addr {
            opts.postgres_options = Some(PostgresOptions {
                addr,
@@ -294,6 +339,7 @@ mod tests {
            http_addr: None,
            rpc_addr: None,
            mysql_addr: None,
+            prom_addr: None,
            postgres_addr: None,
            opentsdb_addr: None,
            config_file: Some(format!(
@@ -339,6 +385,7 @@ mod tests {
        let command = StartCommand {
            http_addr: None,
            rpc_addr: None,
+            prom_addr: None,
            mysql_addr: None,
            postgres_addr: None,
            opentsdb_addr: None,
@@ -362,4 +409,11 @@ mod tests {
            .await;
        assert!(result.is_ok());
    }
+
+    #[test]
+    fn test_toml() {
+        let opts = StandaloneOptions::default();
+        let toml_string = toml::to_string(&opts).unwrap();
+        let _parsed: StandaloneOptions = toml::from_str(&toml_string).unwrap();
+    }
 }
--- a/src/cmd/src/toml_loader.rs
+++ b/src/cmd/src/toml_loader.rs
@@ -29,9 +29,9 @@ mod tests {
    use std::fs::File;
    use std::io::Write;

+    use common_test_util::temp_dir::create_temp_dir;
    use serde::{Deserialize, Serialize};
    use snafu::ResultExt;
-    use tempdir::TempDir;

    use super::*;
    use crate::error::Result;
@@ -62,7 +62,7 @@ mod tests {
            host: "greptime.test".to_string(),
        };

-        let dir = TempDir::new("test_from_file").unwrap();
+        let dir = create_temp_dir("test_from_file");
        let test_file = format!("{}/test.toml", dir.path().to_str().unwrap());

        let s = toml::to_string(&config).unwrap();
--- a/src/cmd/tests/cli.rs
+++ b/src/cmd/tests/cli.rs
@@ -0,0 +1,148 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#[cfg(target_os = "macos")]
+mod tests {
+    use std::path::PathBuf;
+    use std::process::{Command, Stdio};
+    use std::time::Duration;
+
+    use common_test_util::temp_dir::create_temp_dir;
+    use rexpect::session::PtyReplSession;
+
+    struct Repl {
+        repl: PtyReplSession,
+    }
+
+    impl Repl {
+        fn send_line(&mut self, line: &str) {
+            self.repl.send_line(line).unwrap();
+
+            // read a line to consume the prompt
+            self.read_line();
+        }
+
+        fn read_line(&mut self) -> String {
+            self.repl.read_line().unwrap()
+        }
+
+        fn read_expect(&mut self, expect: &str) {
+            assert_eq!(self.read_line(), expect);
+        }
+
+        fn read_contains(&mut self, pat: &str) {
+            assert!(self.read_line().contains(pat));
+        }
+    }
+
+    // TODO(LFC): Un-ignore this REPL test.
+    // Ignore this REPL test because some logical plans like create database are not supported yet in Datanode.
+    #[ignore]
+    #[test]
+    fn test_repl() {
+        let data_dir = create_temp_dir("data");
+        let wal_dir = create_temp_dir("wal");
+
+        let mut bin_path = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+        bin_path.push("../../target/debug");
+        let bin_path = bin_path.to_str().unwrap();
+
+        let mut datanode = Command::new("./greptime")
+            .current_dir(bin_path)
+            .args([
+                "datanode",
+                "start",
+                "--rpc-addr=0.0.0.0:4321",
+                "--node-id=1",
+                &format!("--data-dir={}", data_dir.path().display()),
+                &format!("--wal-dir={}", wal_dir.path().display()),
+            ])
+            .stdout(Stdio::null())
+            .spawn()
+            .unwrap();
+
+        // wait for Datanode actually started
+        std::thread::sleep(Duration::from_secs(3));
+
+        let mut repl_cmd = Command::new("./greptime");
+        repl_cmd.current_dir(bin_path).args([
+            "--log-level=off",
+            "cli",
+            "attach",
+            "--grpc-addr=0.0.0.0:4321",
+            // history commands can sneaky into stdout and mess up our tests, so disable it
+            "--disable-helper",
+        ]);
+        let pty_session = rexpect::session::spawn_command(repl_cmd, Some(5_000)).unwrap();
+        let repl = PtyReplSession {
+            prompt: "> ".to_string(),
+            pty_session,
+            quit_command: None,
+            echo_on: false,
+        };
+        let repl = &mut Repl { repl };
+        repl.read_expect("Ready for commands. (Hint: try 'help')");
+
+        test_create_database(repl);
+
+        test_use_database(repl);
+
+        test_create_table(repl);
+
+        test_insert(repl);
+
+        test_select(repl);
+
+        datanode.kill().unwrap();
+        datanode.wait().unwrap();
+    }
+
+    fn test_create_database(repl: &mut Repl) {
+        repl.send_line("CREATE DATABASE db;");
+        repl.read_expect("Affected Rows: 1");
+        repl.read_contains("Cost");
+    }
+
+    fn test_use_database(repl: &mut Repl) {
+        repl.send_line("USE db");
+        repl.read_expect("Total Rows: 0");
+        repl.read_contains("Cost");
+        repl.read_expect("Using db");
+    }
+
+    fn test_create_table(repl: &mut Repl) {
+        repl.send_line("CREATE TABLE t(x STRING, ts TIMESTAMP TIME INDEX);");
+        repl.read_expect("Affected Rows: 0");
+        repl.read_contains("Cost");
+    }
+
+    fn test_insert(repl: &mut Repl) {
+        repl.send_line("INSERT INTO t(x, ts) VALUES ('hello', 1676895812239);");
+        repl.read_expect("Affected Rows: 1");
+        repl.read_contains("Cost");
+    }
+
+    fn test_select(repl: &mut Repl) {
+        repl.send_line("SELECT * FROM t;");
+
+        repl.read_expect("+-------+-------------------------+");
+        repl.read_expect("| x     | ts                      |");
+        repl.read_expect("+-------+-------------------------+");
+        repl.read_expect("| hello | 2023-02-20T12:23:32.239 |");
+        repl.read_expect("+-------+-------------------------+");
+        repl.read_expect("Total Rows: 1");
+
+        repl.read_contains("Cost");
+    }
+}
--- a/src/common/base/src/buffer.rs
+++ b/src/common/base/src/buffer.rs
@@ -18,7 +18,7 @@ use std::io::{Read, Write};
 use bytes::{Buf, BufMut, BytesMut};
 use common_error::prelude::ErrorExt;
 use paste::paste;
-use snafu::{ensure, Backtrace, ErrorCompat, ResultExt, Snafu};
+use snafu::{ensure, Location, ResultExt, Snafu};

 #[derive(Debug, Snafu)]
 #[snafu(visibility(pub))]
@@ -31,29 +31,33 @@ pub enum Error {
    Overflow {
        src_len: usize,
        dst_len: usize,
-        backtrace: Backtrace,
+        location: Location,
    },

    #[snafu(display("Buffer underflow"))]
-    Underflow { backtrace: Backtrace },
+    Underflow { location: Location },

    #[snafu(display("IO operation reach EOF, source: {}", source))]
    Eof {
        source: std::io::Error,
-        backtrace: Backtrace,
+        location: Location,
    },
 }

 pub type Result<T> = std::result::Result<T, Error>;

 impl ErrorExt for Error {
-    fn backtrace_opt(&self) -> Option<&Backtrace> {
-        ErrorCompat::backtrace(self)
-    }
-
    fn as_any(&self) -> &dyn Any {
        self
    }
+
+    fn location_opt(&self) -> Option<common_error::snafu::Location> {
+        match self {
+            Error::Overflow { location, .. } => Some(*location),
+            Error::Underflow { location, .. } => Some(*location),
+            Error::Eof { location, .. } => Some(*location),
+        }
+    }
 }

 macro_rules! impl_read_le {
--- a/src/common/base/src/bytes.rs
+++ b/src/common/base/src/bytes.rs
@@ -20,6 +20,12 @@ use serde::{Deserialize, Deserializer, Serialize, Serializer};
 #[derive(Debug, Default, Clone, PartialEq, Eq, PartialOrd, Ord, Deserialize, Serialize)]
 pub struct Bytes(bytes::Bytes);

+impl From<Bytes> for bytes::Bytes {
+    fn from(value: Bytes) -> Self {
+        value.0
+    }
+}
+
 impl From<bytes::Bytes> for Bytes {
    fn from(bytes: bytes::Bytes) -> Bytes {
        Bytes(bytes)
--- a/src/common/base/src/readable_size.rs
+++ b/src/common/base/src/readable_size.rs
@@ -53,6 +53,10 @@ impl ReadableSize {
    pub const fn as_mb(self) -> u64 {
        self.0 / MIB
    }
+
+    pub const fn as_bytes(self) -> u64 {
+        self.0
+    }
 }

 impl Div<u64> for ReadableSize {
--- a/src/common/catalog/Cargo.toml
+++ b/src/common/catalog/Cargo.toml
@@ -17,5 +17,4 @@ snafu = { version = "0.7", features = ["backtraces"] }

 [dev-dependencies]
 chrono.workspace = true
-tempdir = "0.3"
 tokio.workspace = true
--- a/src/common/catalog/src/consts.rs
+++ b/src/common/catalog/src/consts.rs
@@ -25,3 +25,6 @@ pub const MIN_USER_TABLE_ID: u32 = 1024;
 pub const SYSTEM_CATALOG_TABLE_ID: u32 = 0;
 /// scripts table id
 pub const SCRIPTS_TABLE_ID: u32 = 1;
+
+pub const MITO_ENGINE: &str = "mito";
+pub const IMMUTABLE_FILE_ENGINE: &str = "file";
--- a/src/common/catalog/src/error.rs
+++ b/src/common/catalog/src/error.rs
@@ -16,29 +16,29 @@ use std::any::Any;

 use common_error::ext::ErrorExt;
 use common_error::prelude::{Snafu, StatusCode};
-use snafu::{Backtrace, ErrorCompat};
+use snafu::Location;

 #[derive(Debug, Snafu)]
 #[snafu(visibility(pub))]
 pub enum Error {
    #[snafu(display("Invalid catalog info: {}", key))]
-    InvalidCatalog { key: String, backtrace: Backtrace },
+    InvalidCatalog { key: String, location: Location },

    #[snafu(display("Failed to deserialize catalog entry value: {}", raw))]
    DeserializeCatalogEntryValue {
        raw: String,
-        backtrace: Backtrace,
+        location: Location,
        source: serde_json::error::Error,
    },

    #[snafu(display("Failed to serialize catalog entry value"))]
    SerializeCatalogEntryValue {
-        backtrace: Backtrace,
+        location: Location,
        source: serde_json::error::Error,
    },

    #[snafu(display("Failed to parse node id: {}", key))]
-    ParseNodeId { key: String, backtrace: Backtrace },
+    ParseNodeId { key: String, location: Location },
 }

 impl ErrorExt for Error {
@@ -51,10 +51,6 @@ impl ErrorExt for Error {
        }
    }

-    fn backtrace_opt(&self) -> Option<&Backtrace> {
-        ErrorCompat::backtrace(self)
-    }
-
    fn as_any(&self) -> &dyn Any {
        self
    }
--- a/src/common/datasource/Cargo.toml
+++ b/src/common/datasource/Cargo.toml
@@ -0,0 +1,28 @@
+[package]
+name = "common-datasource"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+
+[dependencies]
+arrow.workspace = true
+arrow-schema.workspace = true
+async-compression = { version = "0.3", features = [
+    "bzip2",
+    "gzip",
+    "xz",
+    "zstd",
+    "futures-io",
+    "tokio",
+] }
+async-trait.workspace = true
+common-error = { path = "../error" }
+common-runtime = { path = "../runtime" }
+datafusion.workspace = true
+futures.workspace = true
+object-store = { path = "../../object-store" }
+regex = "1.7"
+snafu.workspace = true
+tokio.workspace = true
+tokio-util.workspace = true
+url = "2.3"
--- a/src/common/datasource/src/compression.rs
+++ b/src/common/datasource/src/compression.rs
@@ -0,0 +1,84 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::fmt::Display;
+use std::str::FromStr;
+
+use async_compression::tokio::bufread::{BzDecoder, GzipDecoder, XzDecoder, ZstdDecoder};
+use tokio::io::{AsyncRead, BufReader};
+
+use crate::error::{self, Error, Result};
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum CompressionType {
+    /// Gzip-ed file
+    GZIP,
+    /// Bzip2-ed file
+    BZIP2,
+    /// Xz-ed file (liblzma)
+    XZ,
+    /// Zstd-ed file,
+    ZSTD,
+    /// Uncompressed file
+    UNCOMPRESSED,
+}
+
+impl FromStr for CompressionType {
+    type Err = Error;
+
+    fn from_str(s: &str) -> Result<Self> {
+        let s = s.to_uppercase();
+        match s.as_str() {
+            "GZIP" | "GZ" => Ok(Self::GZIP),
+            "BZIP2" | "BZ2" => Ok(Self::BZIP2),
+            "XZ" => Ok(Self::XZ),
+            "ZST" | "ZSTD" => Ok(Self::ZSTD),
+            "" => Ok(Self::UNCOMPRESSED),
+            _ => error::UnsupportedCompressionTypeSnafu {
+                compression_type: s,
+            }
+            .fail(),
+        }
+    }
+}
+
+impl Display for CompressionType {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(match self {
+            Self::GZIP => "GZIP",
+            Self::BZIP2 => "BZIP2",
+            Self::XZ => "XZ",
+            Self::ZSTD => "ZSTD",
+            Self::UNCOMPRESSED => "",
+        })
+    }
+}
+
+impl CompressionType {
+    pub const fn is_compressed(&self) -> bool {
+        !matches!(self, &Self::UNCOMPRESSED)
+    }
+
+    pub fn convert_async_read<T: AsyncRead + Unpin + Send + 'static>(
+        &self,
+        s: T,
+    ) -> Box<dyn AsyncRead + Unpin + Send> {
+        match self {
+            CompressionType::GZIP => Box::new(GzipDecoder::new(BufReader::new(s))),
+            CompressionType::BZIP2 => Box::new(BzDecoder::new(BufReader::new(s))),
+            CompressionType::XZ => Box::new(XzDecoder::new(BufReader::new(s))),
+            CompressionType::ZSTD => Box::new(ZstdDecoder::new(BufReader::new(s))),
+            CompressionType::UNCOMPRESSED => Box::new(s),
+        }
+    }
+}
--- a/src/common/datasource/src/error.rs
+++ b/src/common/datasource/src/error.rs
@@ -0,0 +1,142 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::any::Any;
+
+use common_error::prelude::*;
+use snafu::Location;
+use url::ParseError;
+
+#[derive(Debug, Snafu)]
+#[snafu(visibility(pub))]
+pub enum Error {
+    #[snafu(display("Unsupported compression type: {}", compression_type))]
+    UnsupportedCompressionType { compression_type: String },
+
+    #[snafu(display("Unsupported backend protocol: {}", protocol))]
+    UnsupportedBackendProtocol { protocol: String },
+
+    #[snafu(display("empty host: {}", url))]
+    EmptyHostPath { url: String },
+
+    #[snafu(display("Invalid path: {}", path))]
+    InvalidPath { path: String },
+
+    #[snafu(display("Invalid url: {}, error :{}", url, source))]
+    InvalidUrl { url: String, source: ParseError },
+
+    #[snafu(display("Failed to decompression, source: {}", source))]
+    Decompression {
+        source: object_store::Error,
+        location: Location,
+    },
+
+    #[snafu(display("Failed to build backend, source: {}", source))]
+    BuildBackend {
+        source: object_store::Error,
+        location: Location,
+    },
+
+    #[snafu(display("Failed to read object from path: {}, source: {}", path, source))]
+    ReadObject {
+        path: String,
+        location: Location,
+        source: object_store::Error,
+    },
+
+    #[snafu(display("Failed to read parquet source: {}", source))]
+    ReadParquetSnafu {
+        location: Location,
+        source: datafusion::parquet::errors::ParquetError,
+    },
+
+    #[snafu(display("Failed to convert parquet to schema: {}", source))]
+    ParquetToSchema {
+        location: Location,
+        source: datafusion::parquet::errors::ParquetError,
+    },
+
+    #[snafu(display("Failed to infer schema from file: {}, source: {}", path, source))]
+    InferSchema {
+        path: String,
+        location: Location,
+        source: arrow_schema::ArrowError,
+    },
+
+    #[snafu(display("Failed to list object in path: {}, source: {}", path, source))]
+    ListObjects {
+        path: String,
+        location: Location,
+        source: object_store::Error,
+    },
+
+    #[snafu(display("Invalid connection: {}", msg))]
+    InvalidConnection { msg: String },
+
+    #[snafu(display("Failed to join handle: {}", source))]
+    JoinHandle {
+        location: Location,
+        source: tokio::task::JoinError,
+    },
+}
+
+pub type Result<T> = std::result::Result<T, Error>;
+
+impl ErrorExt for Error {
+    fn status_code(&self) -> StatusCode {
+        use Error::*;
+        match self {
+            BuildBackend { .. } | ListObjects { .. } | ReadObject { .. } => {
+                StatusCode::StorageUnavailable
+            }
+
+            UnsupportedBackendProtocol { .. }
+            | UnsupportedCompressionType { .. }
+            | InvalidConnection { .. }
+            | InvalidUrl { .. }
+            | EmptyHostPath { .. }
+            | InvalidPath { .. }
+            | InferSchema { .. }
+            | ReadParquetSnafu { .. }
+            | ParquetToSchema { .. } => StatusCode::InvalidArguments,
+
+            Decompression { .. } | JoinHandle { .. } => StatusCode::Unexpected,
+        }
+    }
+
+    fn as_any(&self) -> &dyn Any {
+        self
+    }
+
+    fn location_opt(&self) -> Option<common_error::snafu::Location> {
+        use Error::*;
+        match self {
+            BuildBackend { location, .. } => Some(*location),
+            ReadObject { location, .. } => Some(*location),
+            ListObjects { location, .. } => Some(*location),
+            InferSchema { location, .. } => Some(*location),
+            ReadParquetSnafu { location, .. } => Some(*location),
+            ParquetToSchema { location, .. } => Some(*location),
+            Decompression { location, .. } => Some(*location),
+            JoinHandle { location, .. } => Some(*location),
+
+            UnsupportedBackendProtocol { .. }
+            | EmptyHostPath { .. }
+            | InvalidPath { .. }
+            | InvalidUrl { .. }
+            | InvalidConnection { .. }
+            | UnsupportedCompressionType { .. } => None,
+        }
+    }
+}
--- a/src/common/datasource/src/file_format.rs
+++ b/src/common/datasource/src/file_format.rs
@@ -12,18 +12,19 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-use std::sync::Arc;
+pub mod csv;
+pub mod json;
+pub mod parquet;

-use promql_parser::parser::{Expr, Value};
+pub const DEFAULT_SCHEMA_INFER_MAX_RECORD: usize = 1000;
+
+use arrow::datatypes::SchemaRef;
+use async_trait::async_trait;
+use object_store::ObjectStore;

-use crate::engine::Context;
 use crate::error::Result;

-/// An evaluator evaluates given expressions over given fixed timestamps.
-pub struct Evaluator {}
-
-impl Evaluator {
-    pub fn eval(_ctx: &Context, _expr: &Expr) -> Result<Arc<dyn Value>> {
-        unimplemented!();
-    }
+#[async_trait]
+pub trait FileFormat: Send + Sync + std::fmt::Debug {
+    async fn infer_schema(&self, store: &ObjectStore, path: String) -> Result<SchemaRef>;
 }
--- a/src/common/datasource/src/file_format/csv.rs
+++ b/src/common/datasource/src/file_format/csv.rs
@@ -0,0 +1,158 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::sync::Arc;
+
+use arrow::csv::reader::infer_reader_schema as infer_csv_schema;
+use arrow_schema::SchemaRef;
+use async_trait::async_trait;
+use common_runtime;
+use object_store::ObjectStore;
+use snafu::ResultExt;
+use tokio_util::io::SyncIoBridge;
+
+use crate::compression::CompressionType;
+use crate::error::{self, Result};
+use crate::file_format::{self, FileFormat};
+
+#[derive(Debug)]
+pub struct CsvFormat {
+    pub has_header: bool,
+    pub delimiter: u8,
+    pub schema_infer_max_record: Option<usize>,
+    pub compression_type: CompressionType,
+}
+
+impl Default for CsvFormat {
+    fn default() -> Self {
+        Self {
+            has_header: true,
+            delimiter: b',',
+            schema_infer_max_record: Some(file_format::DEFAULT_SCHEMA_INFER_MAX_RECORD),
+            compression_type: CompressionType::UNCOMPRESSED,
+        }
+    }
+}
+
+#[async_trait]
+impl FileFormat for CsvFormat {
+    async fn infer_schema(&self, store: &ObjectStore, path: String) -> Result<SchemaRef> {
+        let reader = store
+            .reader(&path)
+            .await
+            .context(error::ReadObjectSnafu { path: &path })?;
+
+        let decoded = self.compression_type.convert_async_read(reader);
+
+        let delimiter = self.delimiter;
+        let schema_infer_max_record = self.schema_infer_max_record;
+        let has_header = self.has_header;
+
+        common_runtime::spawn_blocking_read(move || {
+            let reader = SyncIoBridge::new(decoded);
+
+            let (schema, _records_read) =
+                infer_csv_schema(reader, delimiter, schema_infer_max_record, has_header)
+                    .context(error::InferSchemaSnafu { path: &path })?;
+
+            Ok(Arc::new(schema))
+        })
+        .await
+        .context(error::JoinHandleSnafu)?
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use super::*;
+    use crate::file_format::FileFormat;
+    use crate::test_util::{self, format_schema, test_store};
+
+    fn test_data_root() -> String {
+        test_util::get_data_dir("tests/csv").display().to_string()
+    }
+
+    #[tokio::test]
+    async fn infer_schema_basic() {
+        let csv = CsvFormat::default();
+        let store = test_store(&test_data_root());
+        let schema = csv
+            .infer_schema(&store, "simple.csv".to_string())
+            .await
+            .unwrap();
+        let formatted: Vec<_> = format_schema(schema);
+
+        assert_eq!(
+            vec![
+                "c1: Utf8: NULL",
+                "c2: Int64: NULL",
+                "c3: Int64: NULL",
+                "c4: Int64: NULL",
+                "c5: Int64: NULL",
+                "c6: Int64: NULL",
+                "c7: Int64: NULL",
+                "c8: Int64: NULL",
+                "c9: Int64: NULL",
+                "c10: Int64: NULL",
+                "c11: Float64: NULL",
+                "c12: Float64: NULL",
+                "c13: Utf8: NULL"
+            ],
+            formatted,
+        );
+    }
+
+    #[tokio::test]
+    async fn infer_schema_with_limit() {
+        let json = CsvFormat {
+            schema_infer_max_record: Some(3),
+            ..CsvFormat::default()
+        };
+        let store = test_store(&test_data_root());
+        let schema = json
+            .infer_schema(&store, "schema_infer_limit.csv".to_string())
+            .await
+            .unwrap();
+        let formatted: Vec<_> = format_schema(schema);
+
+        assert_eq!(
+            vec![
+                "a: Int64: NULL",
+                "b: Float64: NULL",
+                "c: Int64: NULL",
+                "d: Int64: NULL"
+            ],
+            formatted
+        );
+
+        let json = CsvFormat::default();
+        let store = test_store(&test_data_root());
+        let schema = json
+            .infer_schema(&store, "schema_infer_limit.csv".to_string())
+            .await
+            .unwrap();
+        let formatted: Vec<_> = format_schema(schema);
+
+        assert_eq!(
+            vec![
+                "a: Int64: NULL",
+                "b: Float64: NULL",
+                "c: Int64: NULL",
+                "d: Utf8: NULL"
+            ],
+            formatted
+        );
+    }
+}
--- a/src/common/datasource/src/file_format/json.rs
+++ b/src/common/datasource/src/file_format/json.rs
@@ -0,0 +1,121 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::io::BufReader;
+use std::sync::Arc;
+
+use arrow::datatypes::SchemaRef;
+use arrow::json::reader::{infer_json_schema_from_iterator, ValueIter};
+use async_trait::async_trait;
+use common_runtime;
+use object_store::ObjectStore;
+use snafu::ResultExt;
+use tokio_util::io::SyncIoBridge;
+
+use crate::compression::CompressionType;
+use crate::error::{self, Result};
+use crate::file_format::{self, FileFormat};
+
+#[derive(Debug)]
+pub struct JsonFormat {
+    pub schema_infer_max_record: Option<usize>,
+    pub compression_type: CompressionType,
+}
+
+impl Default for JsonFormat {
+    fn default() -> Self {
+        Self {
+            schema_infer_max_record: Some(file_format::DEFAULT_SCHEMA_INFER_MAX_RECORD),
+            compression_type: CompressionType::UNCOMPRESSED,
+        }
+    }
+}
+
+#[async_trait]
+impl FileFormat for JsonFormat {
+    async fn infer_schema(&self, store: &ObjectStore, path: String) -> Result<SchemaRef> {
+        let reader = store
+            .reader(&path)
+            .await
+            .context(error::ReadObjectSnafu { path: &path })?;
+
+        let decoded = self.compression_type.convert_async_read(reader);
+
+        let schema_infer_max_record = self.schema_infer_max_record;
+
+        common_runtime::spawn_blocking_read(move || {
+            let mut reader = BufReader::new(SyncIoBridge::new(decoded));
+
+            let iter = ValueIter::new(&mut reader, schema_infer_max_record);
+
+            let schema = infer_json_schema_from_iterator(iter)
+                .context(error::InferSchemaSnafu { path: &path })?;
+
+            Ok(Arc::new(schema))
+        })
+        .await
+        .context(error::JoinHandleSnafu)?
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::file_format::FileFormat;
+    use crate::test_util::{self, format_schema, test_store};
+
+    fn test_data_root() -> String {
+        test_util::get_data_dir("tests/json").display().to_string()
+    }
+
+    #[tokio::test]
+    async fn infer_schema_basic() {
+        let json = JsonFormat::default();
+        let store = test_store(&test_data_root());
+        let schema = json
+            .infer_schema(&store, "simple.json".to_string())
+            .await
+            .unwrap();
+        let formatted: Vec<_> = format_schema(schema);
+
+        assert_eq!(
+            vec![
+                "a: Int64: NULL",
+                "b: Float64: NULL",
+                "c: Boolean: NULL",
+                "d: Utf8: NULL",
+            ],
+            formatted
+        );
+    }
+
+    #[tokio::test]
+    async fn infer_schema_with_limit() {
+        let json = JsonFormat {
+            schema_infer_max_record: Some(3),
+            ..JsonFormat::default()
+        };
+        let store = test_store(&test_data_root());
+        let schema = json
+            .infer_schema(&store, "schema_infer_limit.json".to_string())
+            .await
+            .unwrap();
+        let formatted: Vec<_> = format_schema(schema);
+
+        assert_eq!(
+            vec!["a: Int64: NULL", "b: Float64: NULL", "c: Boolean: NULL"],
+            formatted
+        );
+    }
+}
--- a/src/common/datasource/src/file_format/parquet.rs
+++ b/src/common/datasource/src/file_format/parquet.rs
@@ -0,0 +1,78 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::sync::Arc;
+
+use arrow_schema::SchemaRef;
+use async_trait::async_trait;
+use datafusion::parquet::arrow::async_reader::AsyncFileReader;
+use datafusion::parquet::arrow::parquet_to_arrow_schema;
+use object_store::ObjectStore;
+use snafu::ResultExt;
+
+use crate::error::{self, Result};
+use crate::file_format::FileFormat;
+
+#[derive(Debug, Default)]
+pub struct ParquetFormat {}
+
+#[async_trait]
+impl FileFormat for ParquetFormat {
+    async fn infer_schema(&self, store: &ObjectStore, path: String) -> Result<SchemaRef> {
+        let mut reader = store
+            .reader(&path)
+            .await
+            .context(error::ReadObjectSnafu { path: &path })?;
+
+        let metadata = reader
+            .get_metadata()
+            .await
+            .context(error::ReadParquetSnafuSnafu)?;
+
+        let file_metadata = metadata.file_metadata();
+        let schema = parquet_to_arrow_schema(
+            file_metadata.schema_descr(),
+            file_metadata.key_value_metadata(),
+        )
+        .context(error::ParquetToSchemaSnafu)?;
+
+        Ok(Arc::new(schema))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::file_format::FileFormat;
+    use crate::test_util::{self, format_schema, test_store};
+
+    fn test_data_root() -> String {
+        test_util::get_data_dir("tests/parquet")
+            .display()
+            .to_string()
+    }
+
+    #[tokio::test]
+    async fn infer_schema_basic() {
+        let json = ParquetFormat::default();
+        let store = test_store(&test_data_root());
+        let schema = json
+            .infer_schema(&store, "basic.parquet".to_string())
+            .await
+            .unwrap();
+        let formatted: Vec<_> = format_schema(schema);
+
+        assert_eq!(vec!["num: Int64: NULL", "str: Utf8: NULL"], formatted);
+    }
+}
--- a/src/common/datasource/src/lib.rs
+++ b/src/common/datasource/src/lib.rs
@@ -0,0 +1,21 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+pub mod compression;
+pub mod error;
+pub mod file_format;
+pub mod lister;
+pub mod object_store;
+pub mod test_util;
+pub mod util;
--- a/src/common/datasource/src/lister.rs
+++ b/src/common/datasource/src/lister.rs
@@ -0,0 +1,83 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use futures::{future, TryStreamExt};
+use object_store::{Entry, ObjectStore};
+use regex::Regex;
+use snafu::ResultExt;
+
+use crate::error::{self, Result};
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum Source {
+    Filename(String),
+    Dir,
+}
+
+pub struct Lister {
+    object_store: ObjectStore,
+    source: Source,
+    path: String,
+    regex: Option<Regex>,
+}
+
+impl Lister {
+    pub fn new(
+        object_store: ObjectStore,
+        source: Source,
+        path: String,
+        regex: Option<Regex>,
+    ) -> Self {
+        Lister {
+            object_store,
+            source,
+            path,
+            regex,
+        }
+    }
+
+    pub async fn list(&self) -> Result<Vec<Entry>> {
+        match &self.source {
+            Source::Dir => {
+                let streamer = self
+                    .object_store
+                    .list(&self.path)
+                    .await
+                    .context(error::ListObjectsSnafu { path: &self.path })?;
+
+                streamer
+                    .try_filter(|f| {
+                        let res = self
+                            .regex
+                            .as_ref()
+                            .map(|x| x.is_match(f.name()))
+                            .unwrap_or(true);
+                        future::ready(res)
+                    })
+                    .try_collect::<Vec<_>>()
+                    .await
+                    .context(error::ListObjectsSnafu { path: &self.path })
+            }
+            Source::Filename(filename) => {
+                // make sure this file exists
+                let file_full_path = format!("{}{}", self.path, filename);
+                let _ = self.object_store.stat(&file_full_path).await.context(
+                    error::ListObjectsSnafu {
+                        path: &file_full_path,
+                    },
+                )?;
+                Ok(vec![Entry::new(&file_full_path)])
+            }
+        }
+    }
+}
--- a/src/common/datasource/src/object_store.rs
+++ b/src/common/datasource/src/object_store.rs
@@ -0,0 +1,60 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+pub mod fs;
+pub mod s3;
+use std::collections::HashMap;
+
+use object_store::ObjectStore;
+use snafu::{OptionExt, ResultExt};
+use url::{ParseError, Url};
+
+use self::fs::build_fs_backend;
+use self::s3::build_s3_backend;
+use crate::error::{self, Result};
+
+pub const FS_SCHEMA: &str = "FS";
+pub const S3_SCHEMA: &str = "S3";
+
+/// parse url returns (schema,Option<host>,path)
+pub fn parse_url(url: &str) -> Result<(String, Option<String>, String)> {
+    let parsed_url = Url::parse(url);
+    match parsed_url {
+        Ok(url) => Ok((
+            url.scheme().to_string(),
+            url.host_str().map(|s| s.to_string()),
+            url.path().to_string(),
+        )),
+        Err(ParseError::RelativeUrlWithoutBase) => {
+            Ok((FS_SCHEMA.to_string(), None, url.to_string()))
+        }
+        Err(err) => Err(err).context(error::InvalidUrlSnafu { url }),
+    }
+}
+
+pub fn build_backend(url: &str, connection: HashMap<String, String>) -> Result<ObjectStore> {
+    let (schema, host, _path) = parse_url(url)?;
+
+    match schema.to_uppercase().as_str() {
+        S3_SCHEMA => {
+            let host = host.context(error::EmptyHostPathSnafu {
+                url: url.to_string(),
+            })?;
+            Ok(build_s3_backend(&host, "/", connection)?)
+        }
+        FS_SCHEMA => Ok(build_fs_backend("/")?),
+
+        _ => error::UnsupportedBackendProtocolSnafu { protocol: schema }.fail(),
+    }
+}
--- a/src/common/datasource/src/object_store/fs.rs
+++ b/src/common/datasource/src/object_store/fs.rs
@@ -0,0 +1,28 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use object_store::services::Fs;
+use object_store::ObjectStore;
+use snafu::ResultExt;
+
+use crate::error::{BuildBackendSnafu, Result};
+
+pub fn build_fs_backend(root: &str) -> Result<ObjectStore> {
+    let mut builder = Fs::default();
+    builder.root(root);
+    let object_store = ObjectStore::new(builder)
+        .context(BuildBackendSnafu)?
+        .finish();
+    Ok(object_store)
+}
--- a/src/common/datasource/src/object_store/s3.rs
+++ b/src/common/datasource/src/object_store/s3.rs
@@ -0,0 +1,79 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::collections::HashMap;
+
+use object_store::services::S3;
+use object_store::ObjectStore;
+use snafu::ResultExt;
+
+use crate::error::{self, Result};
+
+const ENDPOINT_URL: &str = "ENDPOINT_URL";
+const ACCESS_KEY_ID: &str = "ACCESS_KEY_ID";
+const SECRET_ACCESS_KEY: &str = "SECRET_ACCESS_KEY";
+const SESSION_TOKEN: &str = "SESSION_TOKEN";
+const REGION: &str = "REGION";
+const ENABLE_VIRTUAL_HOST_STYLE: &str = "ENABLE_VIRTUAL_HOST_STYLE";
+
+pub fn build_s3_backend(
+    host: &str,
+    path: &str,
+    connection: HashMap<String, String>,
+) -> Result<ObjectStore> {
+    let mut builder = S3::default();
+
+    builder.root(path);
+
+    builder.bucket(host);
+
+    if let Some(endpoint) = connection.get(ENDPOINT_URL) {
+        builder.endpoint(endpoint);
+    }
+
+    if let Some(region) = connection.get(REGION) {
+        builder.region(region);
+    }
+
+    if let Some(key_id) = connection.get(ACCESS_KEY_ID) {
+        builder.access_key_id(key_id);
+    }
+
+    if let Some(key) = connection.get(SECRET_ACCESS_KEY) {
+        builder.secret_access_key(key);
+    }
+
+    if let Some(session_token) = connection.get(SESSION_TOKEN) {
+        builder.security_token(session_token);
+    }
+
+    if let Some(enable_str) = connection.get(ENABLE_VIRTUAL_HOST_STYLE) {
+        let enable = enable_str.as_str().parse::<bool>().map_err(|e| {
+            error::InvalidConnectionSnafu {
+                msg: format!(
+                    "failed to parse the option {}={}, {}",
+                    ENABLE_VIRTUAL_HOST_STYLE, enable_str, e
+                ),
+            }
+            .build()
+        })?;
+        if enable {
+            builder.enable_virtual_host_style();
+        }
+    }
+
+    Ok(ObjectStore::new(builder)
+        .context(error::BuildBackendSnafu)?
+        .finish())
+}
--- a/src/common/datasource/src/test_util.rs
+++ b/src/common/datasource/src/test_util.rs
@@ -0,0 +1,48 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::path::PathBuf;
+
+use arrow_schema::SchemaRef;
+use object_store::services::Fs;
+use object_store::ObjectStore;
+
+pub fn get_data_dir(path: &str) -> PathBuf {
+    // https://doc.rust-lang.org/cargo/reference/environment-variables.html
+    let dir = env!("CARGO_MANIFEST_DIR");
+
+    PathBuf::from(dir).join(path)
+}
+
+pub fn format_schema(schema: SchemaRef) -> Vec<String> {
+    schema
+        .fields()
+        .iter()
+        .map(|f| {
+            format!(
+                "{}: {:?}: {}",
+                f.name(),
+                f.data_type(),
+                if f.is_nullable() { "NULL" } else { "NOT NULL" }
+            )
+        })
+        .collect()
+}
+
+pub fn test_store(root: &str) -> ObjectStore {
+    let mut builder = Fs::default();
+    builder.root(root);
+
+    ObjectStore::new(builder).unwrap().finish()
+}
--- a/src/common/datasource/src/util.rs
+++ b/src/common/datasource/src/util.rs
@@ -0,0 +1,125 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+pub fn find_dir_and_filename(path: &str) -> (String, Option<String>) {
+    if path.is_empty() {
+        ("/".to_string(), None)
+    } else if path.ends_with('/') {
+        (path.to_string(), None)
+    } else if let Some(idx) = path.rfind('/') {
+        (
+            path[..idx + 1].to_string(),
+            Some(path[idx + 1..].to_string()),
+        )
+    } else {
+        ("/".to_string(), Some(path.to_string()))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use url::Url;
+
+    use super::*;
+
+    #[test]
+    fn test_parse_uri() {
+        struct Test<'a> {
+            uri: &'a str,
+            expected_path: &'a str,
+            expected_schema: &'a str,
+        }
+
+        let tests = [
+            Test {
+                uri: "s3://bucket/to/path/",
+                expected_path: "/to/path/",
+                expected_schema: "s3",
+            },
+            Test {
+                uri: "fs:///to/path/",
+                expected_path: "/to/path/",
+                expected_schema: "fs",
+            },
+            Test {
+                uri: "fs:///to/path/file",
+                expected_path: "/to/path/file",
+                expected_schema: "fs",
+            },
+        ];
+        for test in tests {
+            let parsed_uri = Url::parse(test.uri).unwrap();
+            assert_eq!(parsed_uri.path(), test.expected_path);
+            assert_eq!(parsed_uri.scheme(), test.expected_schema);
+        }
+    }
+
+    #[test]
+    fn test_parse_path_and_dir() {
+        let parsed = Url::from_file_path("/to/path/file").unwrap();
+        assert_eq!(parsed.path(), "/to/path/file");
+
+        let parsed = Url::from_directory_path("/to/path/").unwrap();
+        assert_eq!(parsed.path(), "/to/path/");
+    }
+
+    #[test]
+    fn test_find_dir_and_filename() {
+        struct Test<'a> {
+            path: &'a str,
+            expected_dir: &'a str,
+            expected_filename: Option<String>,
+        }
+
+        let tests = [
+            Test {
+                path: "to/path/",
+                expected_dir: "to/path/",
+                expected_filename: None,
+            },
+            Test {
+                path: "to/path/filename",
+                expected_dir: "to/path/",
+                expected_filename: Some("filename".into()),
+            },
+            Test {
+                path: "/to/path/filename",
+                expected_dir: "/to/path/",
+                expected_filename: Some("filename".into()),
+            },
+            Test {
+                path: "/",
+                expected_dir: "/",
+                expected_filename: None,
+            },
+            Test {
+                path: "filename",
+                expected_dir: "/",
+                expected_filename: Some("filename".into()),
+            },
+            Test {
+                path: "",
+                expected_dir: "/",
+                expected_filename: None,
+            },
+        ];
+
+        for test in tests {
+            let (path, filename) = find_dir_and_filename(test.path);
+            assert_eq!(test.expected_dir, path);
+            assert_eq!(test.expected_filename, filename)
+        }
+    }
+}
--- a/src/common/datasource/tests/README.md
+++ b/src/common/datasource/tests/README.md
@@ -0,0 +1,24 @@
+### Parquet
+The `parquet/basic.parquet` was converted from `csv/basic.csv` via [bdt](https://github.com/andygrove/bdt).
+
+Internal of  `parquet/basic.parquet`: 
+
+Data: 
+```
+-----+-------+
+| num | str   |
+-----+-------+
+| 5   | test  |
+| 2   | hello |
+| 4   | foo   |
+-----+-------+
+```
+Schema:
+```
+-------------+-----------+-------------+
+| column_name | data_type | is_nullable |
+-------------+-----------+-------------+
+| num         | Int64     | YES         |
+| str         | Utf8      | YES         |
+-------------+-----------+-------------+
+```
--- a/src/common/datasource/tests/csv/basic.csv
+++ b/src/common/datasource/tests/csv/basic.csv
@@ -0,0 +1,4 @@
+num,str
+5,test
+2,hello
+4,foo
--- a/src/common/datasource/tests/csv/schema_infer_limit.csv
+++ b/src/common/datasource/tests/csv/schema_infer_limit.csv
@@ -0,0 +1,5 @@
+a,b,c,d
+1,2,3,4
+1,2,3,4
+1,2.0,3,4
+1,2,4,test
--- a/Show More
+++ b/Show More