feat: refetch sst on open

feat: add vec_subvector function (#5683 )
* feat: add vec_subvector function * change datatype of arg1 and arg2 from u64 to i64 * add sqlness test * improve description comments
2025-12-25 23:49:58 +00:00 · 2025-03-17 15:17:14 +08:00 · 2025-03-16 10:43:53 +00:00 · 2025-03-16 09:58:19 +00:00 · 2025-03-16 09:57:46 +00:00 · 2025-03-14 09:31:35 +00:00
179 changed files with 4905 additions and 1935 deletions
--- a/.github/actions/build-linux-artifacts/action.yml
+++ b/.github/actions/build-linux-artifacts/action.yml
@@ -52,7 +52,7 @@ runs:
      uses: ./.github/actions/build-greptime-binary
      with:
        base-image: ubuntu
-        features: servers/dashboard,pg_kvbackend
+        features: servers/dashboard,pg_kvbackend,mysql_kvbackend
        cargo-profile: ${{ inputs.cargo-profile }}
        artifacts-dir: greptime-linux-${{ inputs.arch }}-${{ inputs.version }}
        version: ${{ inputs.version }}
@@ -70,7 +70,7 @@ runs:
      if: ${{ inputs.arch == 'amd64' && inputs.dev-mode == 'false' }} # Builds greptime for centos if the host machine is amd64.
      with:
        base-image: centos
-        features: servers/dashboard,pg_kvbackend
+        features: servers/dashboard,pg_kvbackend,mysql_kvbackend
        cargo-profile: ${{ inputs.cargo-profile }}
        artifacts-dir: greptime-linux-${{ inputs.arch }}-centos-${{ inputs.version }}
        version: ${{ inputs.version }}
--- a/.github/workflows/dev-build.yml
+++ b/.github/workflows/dev-build.yml
@@ -238,6 +238,13 @@ jobs:
          version: ${{ needs.allocate-runners.outputs.version }}
          push-latest-tag: false # Don't push the latest tag to registry.
          dev-mode: true # Only build the standard images.
+          
+      - name: Echo Docker image tag to step summary
+        run: |
+          echo "## Docker Image Tag" >> $GITHUB_STEP_SUMMARY
+          echo "Image Tag: \`${{ needs.allocate-runners.outputs.version }}\`" >> $GITHUB_STEP_SUMMARY
+          echo "Full Image Name: \`docker.io/${{ vars.IMAGE_NAMESPACE }}/${{ vars.DEV_BUILD_IMAGE_NAME }}:${{ needs.allocate-runners.outputs.version }}\`" >> $GITHUB_STEP_SUMMARY
+          echo "Pull Command: \`docker pull docker.io/${{ vars.IMAGE_NAMESPACE }}/${{ vars.DEV_BUILD_IMAGE_NAME }}:${{ needs.allocate-runners.outputs.version }}\`" >> $GITHUB_STEP_SUMMARY

      - name: Set build result
        id: set-build-result
--- a/.github/workflows/develop.yml
+++ b/.github/workflows/develop.yml
@@ -111,7 +111,7 @@ jobs:
      - name: Build greptime binaries
        shell: bash
        # `cargo gc` will invoke `cargo build` with specified args
-        run: cargo gc -- --bin greptime --bin sqlness-runner --features pg_kvbackend
+        run: cargo gc -- --bin greptime --bin sqlness-runner --features "pg_kvbackend,mysql_kvbackend"
      - name: Pack greptime binaries
        shell: bash
        run: |
@@ -270,7 +270,7 @@ jobs:
      - name: Build greptime bianry
        shell: bash
        # `cargo gc` will invoke `cargo build` with specified args
-        run: cargo gc --profile ci -- --bin greptime --features pg_kvbackend
+        run: cargo gc --profile ci -- --bin greptime --features "pg_kvbackend,mysql_kvbackend"
      - name: Pack greptime binary
        shell: bash
        run: |
@@ -687,7 +687,7 @@ jobs:
        working-directory: tests-integration/fixtures
        run: docker compose up -d --wait
      - name: Run nextest cases
-        run: cargo nextest run --workspace -F dashboard -F pg_kvbackend
+        run: cargo nextest run --workspace -F dashboard -F pg_kvbackend -F mysql_kvbackend
        env:
          CARGO_BUILD_RUSTFLAGS: "-C link-arg=-fuse-ld=mold"
          RUST_BACKTRACE: 1
@@ -704,6 +704,7 @@ jobs:
          GT_MINIO_ENDPOINT_URL: http://127.0.0.1:9000
          GT_ETCD_ENDPOINTS: http://127.0.0.1:2379
          GT_POSTGRES_ENDPOINTS: postgres://greptimedb:admin@127.0.0.1:5432/postgres
+          GT_MYSQL_ENDPOINTS: mysql://greptimedb:admin@127.0.0.1:3306/mysql
          GT_KAFKA_ENDPOINTS: 127.0.0.1:9092
          GT_KAFKA_SASL_ENDPOINTS: 127.0.0.1:9093
          UNITTEST_LOG_DIR: "__unittest_logs"
@@ -739,7 +740,7 @@ jobs:
        working-directory: tests-integration/fixtures
        run: docker compose up -d --wait
      - name: Run nextest cases
-        run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info -F dashboard -F pg_kvbackend
+        run: cargo llvm-cov nextest --workspace --lcov --output-path lcov.info -F dashboard -F pg_kvbackend -F mysql_kvbackend
        env:
          CARGO_BUILD_RUSTFLAGS: "-C link-arg=-fuse-ld=mold"
          RUST_BACKTRACE: 1
@@ -755,6 +756,7 @@ jobs:
          GT_MINIO_ENDPOINT_URL: http://127.0.0.1:9000
          GT_ETCD_ENDPOINTS: http://127.0.0.1:2379
          GT_POSTGRES_ENDPOINTS: postgres://greptimedb:admin@127.0.0.1:5432/postgres
+          GT_MYSQL_ENDPOINTS: mysql://greptimedb:admin@127.0.0.1:3306/mysql
          GT_KAFKA_ENDPOINTS: 127.0.0.1:9092
          GT_KAFKA_SASL_ENDPOINTS: 127.0.0.1:9093
          UNITTEST_LOG_DIR: "__unittest_logs"
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -91,7 +91,7 @@ env:
  # The scheduled version is '${{ env.NEXT_RELEASE_VERSION }}-nightly-YYYYMMDD', like v0.2.0-nigthly-20230313;
  NIGHTLY_RELEASE_PREFIX: nightly
  # Note: The NEXT_RELEASE_VERSION should be modified manually by every formal release.
-  NEXT_RELEASE_VERSION: v0.13.0
+  NEXT_RELEASE_VERSION: v0.14.0

 jobs:
  allocate-runners:
--- a/.gitignore
+++ b/.gitignore
@@ -54,3 +54,6 @@ tests-fuzz/corpus/
 # Nix
 .direnv
 .envrc
+
+## default data home
+greptimedb_data
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -185,7 +185,7 @@ checksum = "d301b3b94cb4b2f23d7917810addbbaff90738e0ca2be692bd027e70d7e0330c"

 [[package]]
 name = "api"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "common-base",
 "common-decimal",
@@ -710,7 +710,7 @@ dependencies = [

 [[package]]
 name = "auth"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "async-trait",
@@ -1324,7 +1324,7 @@ dependencies = [

 [[package]]
 name = "cache"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "catalog",
 "common-error",
@@ -1348,7 +1348,7 @@ checksum = "37b2a672a2cb129a2e41c10b1224bb368f9f37a2b16b612598138befd7b37eb5"

 [[package]]
 name = "catalog"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arrow",
@@ -1661,7 +1661,7 @@ checksum = "1462739cb27611015575c0c11df5df7601141071f07518d56fcc1be504cbec97"

 [[package]]
 name = "cli"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "auth",
@@ -1693,6 +1693,7 @@ dependencies = [
 "humantime",
 "meta-client",
 "nu-ansi-term",
+ "opendal",
 "query",
 "rand",
 "reqwest",
@@ -1703,7 +1704,7 @@ dependencies = [
 "session",
 "snafu 0.8.5",
 "store-api",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "table",
 "tempfile",
 "tokio",
@@ -1712,7 +1713,7 @@ dependencies = [

 [[package]]
 name = "client"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arc-swap",
@@ -1739,7 +1740,7 @@ dependencies = [
 "rand",
 "serde_json",
 "snafu 0.8.5",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "substrait 0.37.3",
 "tokio",
 "tokio-stream",
@@ -1780,7 +1781,7 @@ dependencies = [

 [[package]]
 name = "cmd"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "auth",
@@ -1841,7 +1842,7 @@ dependencies = [
 "similar-asserts",
 "snafu 0.8.5",
 "store-api",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "table",
 "temp-env",
 "tempfile",
@@ -1887,7 +1888,7 @@ checksum = "55b672471b4e9f9e95499ea597ff64941a309b2cdbffcc46f2cc5e2d971fd335"

 [[package]]
 name = "common-base"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "anymap2",
 "async-trait",
@@ -1909,11 +1910,11 @@ dependencies = [

 [[package]]
 name = "common-catalog"
-version = "0.13.0"
+version = "0.14.0"

 [[package]]
 name = "common-config"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "common-base",
 "common-error",
@@ -1938,7 +1939,7 @@ dependencies = [

 [[package]]
 name = "common-datasource"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "arrow",
 "arrow-schema",
@@ -1974,7 +1975,7 @@ dependencies = [

 [[package]]
 name = "common-decimal"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "bigdecimal 0.4.5",
 "common-error",
@@ -1987,7 +1988,7 @@ dependencies = [

 [[package]]
 name = "common-error"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "http 1.1.0",
 "snafu 0.8.5",
@@ -1997,7 +1998,7 @@ dependencies = [

 [[package]]
 name = "common-frontend"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "common-error",
@@ -2007,7 +2008,7 @@ dependencies = [

 [[package]]
 name = "common-function"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "ahash 0.8.11",
 "api",
@@ -2015,6 +2016,7 @@ dependencies = [
 "arc-swap",
 "async-trait",
 "bincode",
+ "chrono",
 "common-base",
 "common-catalog",
 "common-error",
@@ -2057,7 +2059,7 @@ dependencies = [

 [[package]]
 name = "common-greptimedb-telemetry"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "common-runtime",
@@ -2074,7 +2076,7 @@ dependencies = [

 [[package]]
 name = "common-grpc"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arrow-flight",
@@ -2102,7 +2104,7 @@ dependencies = [

 [[package]]
 name = "common-grpc-expr"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "common-base",
@@ -2121,7 +2123,7 @@ dependencies = [

 [[package]]
 name = "common-macro"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "arc-swap",
 "common-query",
@@ -2135,7 +2137,7 @@ dependencies = [

 [[package]]
 name = "common-mem-prof"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "common-error",
 "common-macro",
@@ -2148,7 +2150,7 @@ dependencies = [

 [[package]]
 name = "common-meta"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "anymap2",
 "api",
@@ -2196,6 +2198,7 @@ dependencies = [
 "serde_with",
 "session",
 "snafu 0.8.5",
+ "sqlx",
 "store-api",
 "strum 0.25.0",
 "table",
@@ -2208,7 +2211,7 @@ dependencies = [

 [[package]]
 name = "common-options"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "common-grpc",
 "humantime-serde",
@@ -2217,11 +2220,11 @@ dependencies = [

 [[package]]
 name = "common-plugins"
-version = "0.13.0"
+version = "0.14.0"

 [[package]]
 name = "common-pprof"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "common-error",
 "common-macro",
@@ -2233,7 +2236,7 @@ dependencies = [

 [[package]]
 name = "common-procedure"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-stream",
 "async-trait",
@@ -2260,7 +2263,7 @@ dependencies = [

 [[package]]
 name = "common-procedure-test"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "common-procedure",
@@ -2268,7 +2271,7 @@ dependencies = [

 [[package]]
 name = "common-query"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "async-trait",
@@ -2294,7 +2297,7 @@ dependencies = [

 [[package]]
 name = "common-recordbatch"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "arc-swap",
 "common-error",
@@ -2313,7 +2316,7 @@ dependencies = [

 [[package]]
 name = "common-runtime"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "clap 4.5.19",
@@ -2343,7 +2346,7 @@ dependencies = [

 [[package]]
 name = "common-telemetry"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "atty",
 "backtrace",
@@ -2371,7 +2374,7 @@ dependencies = [

 [[package]]
 name = "common-test-util"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "client",
 "common-query",
@@ -2383,7 +2386,7 @@ dependencies = [

 [[package]]
 name = "common-time"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "arrow",
 "chrono",
@@ -2401,7 +2404,7 @@ dependencies = [

 [[package]]
 name = "common-version"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "build-data",
 "const_format",
@@ -2411,7 +2414,7 @@ dependencies = [

 [[package]]
 name = "common-wal"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "common-base",
 "common-error",
@@ -3342,7 +3345,7 @@ dependencies = [

 [[package]]
 name = "datanode"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arrow-flight",
@@ -3394,7 +3397,7 @@ dependencies = [
 "session",
 "snafu 0.8.5",
 "store-api",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "table",
 "tokio",
 "toml 0.8.19",
@@ -3403,7 +3406,7 @@ dependencies = [

 [[package]]
 name = "datatypes"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -4047,7 +4050,7 @@ dependencies = [

 [[package]]
 name = "file-engine"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "async-trait",
@@ -4116,11 +4119,12 @@ dependencies = [

 [[package]]
 name = "flate2"
-version = "1.0.34"
+version = "1.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a1b589b4dc103969ad3cf85c950899926ec64300a1a46d76c03a6072957036f0"
+checksum = "11faaf5a5236997af9848be0bef4db95824b1d534ebc64d0f0c6cf3e67bd38dc"
 dependencies = [
 "crc32fast",
+ "libz-rs-sys",
 "libz-sys",
 "miniz_oxide",
 ]
@@ -4157,7 +4161,7 @@ checksum = "8bf7cc16383c4b8d58b9905a8509f02926ce3058053c056376248d958c9df1e8"

 [[package]]
 name = "flow"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arrow",
@@ -4167,6 +4171,7 @@ dependencies = [
 "bytes",
 "cache",
 "catalog",
+ "chrono",
 "client",
 "common-base",
 "common-catalog",
@@ -4218,7 +4223,7 @@ dependencies = [
 "snafu 0.8.5",
 "store-api",
 "strum 0.25.0",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "table",
 "tokio",
 "tonic 0.12.3",
@@ -4273,7 +4278,7 @@ checksum = "6c2141d6d6c8512188a7891b4b01590a45f6dac67afb4f255c4124dbb86d4eaa"

 [[package]]
 name = "frontend"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arc-swap",
@@ -4701,7 +4706,7 @@ dependencies = [
 [[package]]
 name = "greptime-proto"
 version = "0.1.0"
-source = "git+https://github.com/GreptimeTeam/greptime-proto.git?rev=d92c9ac4e90ef4abdcf5c2eaf5a164e18ba09486#d92c9ac4e90ef4abdcf5c2eaf5a164e18ba09486"
+source = "git+https://github.com/GreptimeTeam/greptime-proto.git?rev=c5419bbd20cb42e568ec325a4d71a3c94cc327e1#c5419bbd20cb42e568ec325a4d71a3c94cc327e1"
 dependencies = [
 "prost 0.13.3",
 "serde",
@@ -5541,7 +5546,7 @@ dependencies = [

 [[package]]
 name = "index"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "asynchronous-codec",
@@ -6274,6 +6279,15 @@ dependencies = [
 "vcpkg",
 ]

+[[package]]
+name = "libz-rs-sys"
+version = "0.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "902bc563b5d65ad9bba616b490842ef0651066a1a1dc3ce1087113ffcb873c8d"
+dependencies = [
+ "zlib-rs",
+]
+
 [[package]]
 name = "libz-sys"
 version = "1.1.20"
@@ -6334,7 +6348,7 @@ checksum = "a7a70ba024b9dc04c27ea2f0c0548feb474ec5c54bba33a7f72f873a39d07b24"

 [[package]]
 name = "log-query"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "chrono",
 "common-error",
@@ -6346,7 +6360,7 @@ dependencies = [

 [[package]]
 name = "log-store"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-stream",
 "async-trait",
@@ -6639,7 +6653,7 @@ dependencies = [

 [[package]]
 name = "meta-client"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "async-trait",
@@ -6666,7 +6680,7 @@ dependencies = [

 [[package]]
 name = "meta-srv"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "async-trait",
@@ -6717,6 +6731,7 @@ dependencies = [
 "session",
 "snafu 0.8.5",
 "store-api",
+ "strum 0.25.0",
 "table",
 "tokio",
 "tokio-postgres",
@@ -6752,7 +6767,7 @@ dependencies = [

 [[package]]
 name = "metric-engine"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "aquamarine",
@@ -6817,9 +6832,9 @@ checksum = "68354c5c6bd36d73ff3feceb05efa59b6acb7626617f4962be322a825e61f79a"

 [[package]]
 name = "miniz_oxide"
-version = "0.8.0"
+version = "0.8.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e2d80299ef12ff69b16a84bb182e3b9df68b5a91574d3d4fa6e41b65deec4df1"
+checksum = "8e3e04debbb59698c15bacbb6d93584a8c0ca9cc3213cb423d31f760d8843ce5"
 dependencies = [
 "adler2",
 ]
@@ -6850,7 +6865,7 @@ dependencies = [

 [[package]]
 name = "mito2"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "aquamarine",
@@ -7547,7 +7562,7 @@ dependencies = [

 [[package]]
 name = "object-store"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "anyhow",
 "bytes",
@@ -7796,7 +7811,7 @@ dependencies = [

 [[package]]
 name = "operator"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "ahash 0.8.11",
 "api",
@@ -7844,7 +7859,7 @@ dependencies = [
 "sql",
 "sqlparser 0.52.0 (git+https://github.com/GreptimeTeam/sqlparser-rs.git?rev=71dd86058d2af97b9925093d40c4e03360403170)",
 "store-api",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "table",
 "tokio",
 "tokio-util",
@@ -8081,7 +8096,7 @@ dependencies = [

 [[package]]
 name = "partition"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "async-trait",
@@ -8349,7 +8364,7 @@ checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184"

 [[package]]
 name = "pipeline"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "ahash 0.8.11",
 "api",
@@ -8489,7 +8504,7 @@ dependencies = [

 [[package]]
 name = "plugins"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "auth",
 "clap 4.5.19",
@@ -8751,7 +8766,7 @@ dependencies = [

 [[package]]
 name = "promql"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "ahash 0.8.11",
 "async-trait",
@@ -8998,7 +9013,7 @@ dependencies = [

 [[package]]
 name = "puffin"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-compression 0.4.13",
 "async-trait",
@@ -9039,7 +9054,7 @@ dependencies = [

 [[package]]
 name = "query"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "ahash 0.8.11",
 "api",
@@ -9104,7 +9119,7 @@ dependencies = [
 "sqlparser 0.52.0 (git+https://github.com/GreptimeTeam/sqlparser-rs.git?rev=71dd86058d2af97b9925093d40c4e03360403170)",
 "statrs",
 "store-api",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "table",
 "tokio",
 "tokio-stream",
@@ -10459,7 +10474,7 @@ dependencies = [

 [[package]]
 name = "servers"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "ahash 0.8.11",
 "api",
@@ -10576,7 +10591,7 @@ dependencies = [

 [[package]]
 name = "session"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arc-swap",
@@ -10885,7 +10900,7 @@ dependencies = [

 [[package]]
 name = "sql"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "chrono",
@@ -10939,7 +10954,7 @@ dependencies = [

 [[package]]
 name = "sqlness-runner"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "clap 4.5.19",
@@ -11256,7 +11271,7 @@ dependencies = [

 [[package]]
 name = "store-api"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "aquamarine",
@@ -11386,7 +11401,7 @@ dependencies = [

 [[package]]
 name = "substrait"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "async-trait",
 "bytes",
@@ -11567,7 +11582,7 @@ dependencies = [

 [[package]]
 name = "table"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "async-trait",
@@ -11818,7 +11833,7 @@ checksum = "3369f5ac52d5eb6ab48c6b4ffdc8efbcad6b89c765749064ba298f2c68a16a76"

 [[package]]
 name = "tests-fuzz"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "arbitrary",
 "async-trait",
@@ -11862,7 +11877,7 @@ dependencies = [

 [[package]]
 name = "tests-integration"
-version = "0.13.0"
+version = "0.14.0"
 dependencies = [
 "api",
 "arrow-flight",
@@ -11928,7 +11943,7 @@ dependencies = [
 "sql",
 "sqlx",
 "store-api",
- "substrait 0.13.0",
+ "substrait 0.14.0",
 "table",
 "tempfile",
 "time",
@@ -13949,6 +13964,12 @@ dependencies = [
 "syn 2.0.96",
 ]

+[[package]]
+name = "zlib-rs"
+version = "0.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8b20717f0917c908dc63de2e44e97f1e6b126ca58d0e391cee86d504eb8fbd05"
+
 [[package]]
 name = "zstd"
 version = "0.11.2+zstd.1.5.2"
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -67,7 +67,7 @@ members = [
 resolver = "2"

 [workspace.package]
-version = "0.13.0"
+version = "0.14.0"
 edition = "2021"
 license = "Apache-2.0"

@@ -126,10 +126,11 @@ deadpool-postgres = "0.12"
 derive_builder = "0.12"
 dotenv = "0.15"
 etcd-client = "0.14"
+flate2 = { version = "1.1.0", default-features = false, features = ["zlib-rs"] }
 fst = "0.4.7"
 futures = "0.3"
 futures-util = "0.3"
-greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "d92c9ac4e90ef4abdcf5c2eaf5a164e18ba09486" }
+greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "c5419bbd20cb42e568ec325a4d71a3c94cc327e1" }
 hex = "0.4"
 http = "1"
 humantime = "2.1"
@@ -188,6 +189,10 @@ shadow-rs = "0.38"
 similar-asserts = "1.6.0"
 smallvec = { version = "1", features = ["serde"] }
 snafu = "0.8"
+sqlx = { version = "0.8", features = [
+    "runtime-tokio-rustls",
+    "mysql",
+] }
 sysinfo = "0.30"
 # on branch v0.52.x
 sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "71dd86058d2af97b9925093d40c4e03360403170", features = [
--- a/README.md
+++ b/README.md
@@ -112,7 +112,7 @@ Start a GreptimeDB container with:

 ```shell
 docker run -p 127.0.0.1:4000-4003:4000-4003 \
-  -v "$(pwd)/greptimedb:/tmp/greptimedb" \
+  -v "$(pwd)/greptimedb:./greptimedb_data" \
  --name greptime --rm \
  greptime/greptimedb:latest standalone start \
  --http-addr 0.0.0.0:4000 \
--- a/config/config.md
+++ b/config/config.md
@@ -101,7 +101,7 @@
 | `flow` | -- | -- | flow engine options. |
 | `flow.num_workers` | Integer | `0` | The number of flow worker in flownode.<br/>Not setting(or set to 0) this value will use the number of CPU cores divided by 2. |
 | `storage` | -- | -- | The data storage options. |
-| `storage.data_home` | String | `/tmp/greptimedb/` | The working home directory. |
+| `storage.data_home` | String | `./greptimedb_data/` | The working home directory. |
 | `storage.type` | String | `File` | The storage type used to store the data.<br/>- `File`: the data is stored in the local file system.<br/>- `S3`: the data is stored in the S3 object storage.<br/>- `Gcs`: the data is stored in the Google Cloud Storage.<br/>- `Azblob`: the data is stored in the Azure Blob Storage.<br/>- `Oss`: the data is stored in the Aliyun OSS. |
 | `storage.cache_path` | String | Unset | Read cache configuration for object storage such as 'S3' etc, it's configured by default when using object storage. It is recommended to configure it when using object storage for better performance.<br/>A local file directory, defaults to `{data_home}`. An empty string means disabling. |
 | `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger. |
@@ -181,7 +181,7 @@
 | `region_engine.metric` | -- | -- | Metric engine options. |
 | `region_engine.metric.experimental_sparse_primary_key_encoding` | Bool | `false` | Whether to enable the experimental sparse primary key encoding. |
 | `logging` | -- | -- | The logging options. |
-| `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.dir` | String | `./greptimedb_data/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
 | `logging.level` | String | Unset | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
 | `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
@@ -279,7 +279,7 @@
 | `datanode.client.connect_timeout` | String | `10s` | -- |
 | `datanode.client.tcp_nodelay` | Bool | `true` | -- |
 | `logging` | -- | -- | The logging options. |
-| `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.dir` | String | `./greptimedb_data/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
 | `logging.level` | String | Unset | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
 | `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
@@ -308,7 +308,7 @@

 | Key | Type | Default | Descriptions |
 | --- | -----| ------- | ----------- |
-| `data_home` | String | `/tmp/metasrv/` | The working home directory. |
+| `data_home` | String | `./greptimedb_data/metasrv/` | The working home directory. |
 | `bind_addr` | String | `127.0.0.1:3002` | The bind address of metasrv. |
 | `server_addr` | String | `127.0.0.1:3002` | The communication server address for the frontend and datanode to connect to metasrv.<br/>If left empty or unset, the server will automatically use the IP address of the first network interface<br/>on the host, with the same port number as the one specified in `bind_addr`. |
 | `store_addrs` | Array | -- | Store server address default to etcd store.<br/>For postgres store, the format is:<br/>"password=password dbname=postgres user=postgres host=localhost port=5432"<br/>For etcd store, the format is:<br/>"127.0.0.1:2379" |
@@ -352,7 +352,7 @@
 | `wal.backoff_base` | Integer | `2` | Exponential backoff rate, i.e. next backoff = base * current backoff. |
 | `wal.backoff_deadline` | String | `5mins` | Stop reconnecting if the total wait time reaches the deadline. If this config is missing, the reconnecting won't terminate. |
 | `logging` | -- | -- | The logging options. |
-| `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.dir` | String | `./greptimedb_data/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
 | `logging.level` | String | Unset | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
 | `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
@@ -442,7 +442,7 @@
 | `wal.dump_index_interval` | String | `60s` | The interval for dumping WAL indexes.<br/>**It's only used when the provider is `kafka`**. |
 | `wal.overwrite_entry_start_id` | Bool | `false` | Ignore missing entries during read WAL.<br/>**It's only used when the provider is `kafka`**.<br/><br/>This option ensures that when Kafka messages are deleted, the system<br/>can still successfully replay memtable data without throwing an<br/>out-of-range error.<br/>However, enabling this option might lead to unexpected data loss,<br/>as the system will skip over missing entries instead of treating<br/>them as critical errors. |
 | `storage` | -- | -- | The data storage options. |
-| `storage.data_home` | String | `/tmp/greptimedb/` | The working home directory. |
+| `storage.data_home` | String | `./greptimedb_data/` | The working home directory. |
 | `storage.type` | String | `File` | The storage type used to store the data.<br/>- `File`: the data is stored in the local file system.<br/>- `S3`: the data is stored in the S3 object storage.<br/>- `Gcs`: the data is stored in the Google Cloud Storage.<br/>- `Azblob`: the data is stored in the Azure Blob Storage.<br/>- `Oss`: the data is stored in the Aliyun OSS. |
 | `storage.cache_path` | String | Unset | Read cache configuration for object storage such as 'S3' etc, it's configured by default when using object storage. It is recommended to configure it when using object storage for better performance.<br/>A local file directory, defaults to `{data_home}`. An empty string means disabling. |
 | `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger. |
@@ -522,7 +522,7 @@
 | `region_engine.metric` | -- | -- | Metric engine options. |
 | `region_engine.metric.experimental_sparse_primary_key_encoding` | Bool | `false` | Whether to enable the experimental sparse primary key encoding. |
 | `logging` | -- | -- | The logging options. |
-| `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.dir` | String | `./greptimedb_data/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
 | `logging.level` | String | Unset | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
 | `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
@@ -579,7 +579,7 @@
 | `heartbeat.interval` | String | `3s` | Interval for sending heartbeat messages to the metasrv. |
 | `heartbeat.retry_interval` | String | `3s` | Interval for retrying to send heartbeat messages to the metasrv. |
 | `logging` | -- | -- | The logging options. |
-| `logging.dir` | String | `/tmp/greptimedb/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
+| `logging.dir` | String | `./greptimedb_data/logs` | The directory to store the log files. If set to empty, logs will not be written to files. |
 | `logging.level` | String | Unset | The log level. Can be `info`/`debug`/`warn`/`error`. |
 | `logging.enable_otlp_tracing` | Bool | `false` | Enable OTLP tracing. |
 | `logging.otlp_endpoint` | String | `http://localhost:4317` | The OTLP tracing endpoint. |
--- a/config/datanode.example.toml
+++ b/config/datanode.example.toml
@@ -119,7 +119,7 @@ provider = "raft_engine"
 ## The directory to store the WAL files.
 ## **It's only used when the provider is `raft_engine`**.
 ## @toml2docs:none-default
-dir = "/tmp/greptimedb/wal"
+dir = "./greptimedb_data/wal"

 ## The size of the WAL segment file.
 ## **It's only used when the provider is `raft_engine`**.
@@ -231,6 +231,7 @@ overwrite_entry_start_id = false
 # secret_access_key = "123456"
 # endpoint = "https://s3.amazonaws.com"
 # region = "us-west-2"
+# enable_virtual_host_style = false

 # Example of using Oss as the storage.
 # [storage]
@@ -264,7 +265,7 @@ overwrite_entry_start_id = false
 ## The data storage options.
 [storage]
 ## The working home directory.
-data_home = "/tmp/greptimedb/"
+data_home = "./greptimedb_data/"

 ## The storage type used to store the data.
 ## - `File`: the data is stored in the local file system.
@@ -617,7 +618,7 @@ experimental_sparse_primary_key_encoding = false
 ## The logging options.
 [logging]
 ## The directory to store the log files. If set to empty, logs will not be written to files.
-dir = "/tmp/greptimedb/logs"
+dir = "./greptimedb_data/logs"

 ## The log level. Can be `info`/`debug`/`warn`/`error`.
 ## @toml2docs:none-default
--- a/config/flownode.example.toml
+++ b/config/flownode.example.toml
@@ -76,7 +76,7 @@ retry_interval = "3s"
 ## The logging options.
 [logging]
 ## The directory to store the log files. If set to empty, logs will not be written to files.
-dir = "/tmp/greptimedb/logs"
+dir = "./greptimedb_data/logs"

 ## The log level. Can be `info`/`debug`/`warn`/`error`.
 ## @toml2docs:none-default
@@ -121,4 +121,3 @@ sample_ratio = 1.0
 ## The tokio console address.
 ## @toml2docs:none-default
 #+ tokio_console_addr = "127.0.0.1"
-
--- a/config/frontend.example.toml
+++ b/config/frontend.example.toml
@@ -189,7 +189,7 @@ tcp_nodelay = true
 ## The logging options.
 [logging]
 ## The directory to store the log files. If set to empty, logs will not be written to files.
-dir = "/tmp/greptimedb/logs"
+dir = "./greptimedb_data/logs"

 ## The log level. Can be `info`/`debug`/`warn`/`error`.
 ## @toml2docs:none-default
--- a/config/metasrv.example.toml
+++ b/config/metasrv.example.toml
@@ -1,5 +1,5 @@
 ## The working home directory.
-data_home = "/tmp/metasrv/"
+data_home = "./greptimedb_data/metasrv/"

 ## The bind address of metasrv.
 bind_addr = "127.0.0.1:3002"
@@ -177,7 +177,7 @@ backoff_deadline = "5mins"
 ## The logging options.
 [logging]
 ## The directory to store the log files. If set to empty, logs will not be written to files.
-dir = "/tmp/greptimedb/logs"
+dir = "./greptimedb_data/logs"

 ## The log level. Can be `info`/`debug`/`warn`/`error`.
 ## @toml2docs:none-default
--- a/config/standalone.example.toml
+++ b/config/standalone.example.toml
@@ -164,7 +164,7 @@ provider = "raft_engine"
 ## The directory to store the WAL files.
 ## **It's only used when the provider is `raft_engine`**.
 ## @toml2docs:none-default
-dir = "/tmp/greptimedb/wal"
+dir = "./greptimedb_data/wal"

 ## The size of the WAL segment file.
 ## **It's only used when the provider is `raft_engine`**.
@@ -318,6 +318,7 @@ retry_delay = "500ms"
 # secret_access_key = "123456"
 # endpoint = "https://s3.amazonaws.com"
 # region = "us-west-2"
+# enable_virtual_host_style = false

 # Example of using Oss as the storage.
 # [storage]
@@ -351,7 +352,7 @@ retry_delay = "500ms"
 ## The data storage options.
 [storage]
 ## The working home directory.
-data_home = "/tmp/greptimedb/"
+data_home = "./greptimedb_data/"

 ## The storage type used to store the data.
 ## - `File`: the data is stored in the local file system.
@@ -704,7 +705,7 @@ experimental_sparse_primary_key_encoding = false
 ## The logging options.
 [logging]
 ## The directory to store the log files. If set to empty, logs will not be written to files.
-dir = "/tmp/greptimedb/logs"
+dir = "./greptimedb_data/logs"

 ## The log level. Can be `info`/`debug`/`warn`/`error`.
 ## @toml2docs:none-default
--- a/docker/docker-compose/cluster-with-etcd.yaml
+++ b/docker/docker-compose/cluster-with-etcd.yaml
@@ -25,7 +25,7 @@ services:
      - --initial-cluster-state=new
      - *etcd_initial_cluster_token
    volumes:
-      - /tmp/greptimedb-cluster-docker-compose/etcd0:/var/lib/etcd
+      - ./greptimedb-cluster-docker-compose/etcd0:/var/lib/etcd
    healthcheck:
      test: [ "CMD", "etcdctl", "--endpoints=http://etcd0:2379", "endpoint", "health" ]
      interval: 5s
@@ -68,12 +68,13 @@ services:
      - datanode
      - start
      - --node-id=0
+      - --data-home=/greptimedb_data
      - --rpc-bind-addr=0.0.0.0:3001
      - --rpc-server-addr=datanode0:3001
      - --metasrv-addrs=metasrv:3002
      - --http-addr=0.0.0.0:5000
    volumes:
-      - /tmp/greptimedb-cluster-docker-compose/datanode0:/tmp/greptimedb
+      - ./greptimedb-cluster-docker-compose/datanode0:/greptimedb_data
    healthcheck:
      test: [ "CMD", "curl", "-fv", "http://datanode0:5000/health" ]
      interval: 5s
--- a/src/api/src/helper.rs
+++ b/src/api/src/helper.rs
@@ -19,9 +19,7 @@ use common_decimal::decimal128::{DECIMAL128_DEFAULT_SCALE, DECIMAL128_MAX_PRECIS
 use common_decimal::Decimal128;
 use common_time::time::Time;
 use common_time::timestamp::TimeUnit;
-use common_time::{
-    Date, DateTime, IntervalDayTime, IntervalMonthDayNano, IntervalYearMonth, Timestamp,
-};
+use common_time::{Date, IntervalDayTime, IntervalMonthDayNano, IntervalYearMonth, Timestamp};
 use datatypes::prelude::{ConcreteDataType, ValueRef};
 use datatypes::scalars::ScalarVector;
 use datatypes::types::{
@@ -29,8 +27,8 @@ use datatypes::types::{
 };
 use datatypes::value::{OrderedF32, OrderedF64, Value};
 use datatypes::vectors::{
-    BinaryVector, BooleanVector, DateTimeVector, DateVector, Decimal128Vector, Float32Vector,
-    Float64Vector, Int32Vector, Int64Vector, IntervalDayTimeVector, IntervalMonthDayNanoVector,
+    BinaryVector, BooleanVector, DateVector, Decimal128Vector, Float32Vector, Float64Vector,
+    Int32Vector, Int64Vector, IntervalDayTimeVector, IntervalMonthDayNanoVector,
    IntervalYearMonthVector, PrimitiveVector, StringVector, TimeMicrosecondVector,
    TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector,
    TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, UInt32Vector,
@@ -118,7 +116,7 @@ impl From<ColumnDataTypeWrapper> for ConcreteDataType {
            ColumnDataType::Json => ConcreteDataType::json_datatype(),
            ColumnDataType::String => ConcreteDataType::string_datatype(),
            ColumnDataType::Date => ConcreteDataType::date_datatype(),
-            ColumnDataType::Datetime => ConcreteDataType::datetime_datatype(),
+            ColumnDataType::Datetime => ConcreteDataType::timestamp_microsecond_datatype(),
            ColumnDataType::TimestampSecond => ConcreteDataType::timestamp_second_datatype(),
            ColumnDataType::TimestampMillisecond => {
                ConcreteDataType::timestamp_millisecond_datatype()
@@ -271,7 +269,6 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
            ConcreteDataType::Binary(_) => ColumnDataType::Binary,
            ConcreteDataType::String(_) => ColumnDataType::String,
            ConcreteDataType::Date(_) => ColumnDataType::Date,
-            ConcreteDataType::DateTime(_) => ColumnDataType::Datetime,
            ConcreteDataType::Timestamp(t) => match t {
                TimestampType::Second(_) => ColumnDataType::TimestampSecond,
                TimestampType::Millisecond(_) => ColumnDataType::TimestampMillisecond,
@@ -476,7 +473,6 @@ pub fn push_vals(column: &mut Column, origin_count: usize, vector: VectorRef) {
        Value::String(val) => values.string_values.push(val.as_utf8().to_string()),
        Value::Binary(val) => values.binary_values.push(val.to_vec()),
        Value::Date(val) => values.date_values.push(val.val()),
-        Value::DateTime(val) => values.datetime_values.push(val.val()),
        Value::Timestamp(val) => match val.unit() {
            TimeUnit::Second => values.timestamp_second_values.push(val.value()),
            TimeUnit::Millisecond => values.timestamp_millisecond_values.push(val.value()),
@@ -577,12 +573,11 @@ pub fn pb_value_to_value_ref<'a>(
        ValueData::BinaryValue(bytes) => ValueRef::Binary(bytes.as_slice()),
        ValueData::StringValue(string) => ValueRef::String(string.as_str()),
        ValueData::DateValue(d) => ValueRef::Date(Date::from(*d)),
-        ValueData::DatetimeValue(d) => ValueRef::DateTime(DateTime::new(*d)),
        ValueData::TimestampSecondValue(t) => ValueRef::Timestamp(Timestamp::new_second(*t)),
        ValueData::TimestampMillisecondValue(t) => {
            ValueRef::Timestamp(Timestamp::new_millisecond(*t))
        }
-        ValueData::TimestampMicrosecondValue(t) => {
+        ValueData::DatetimeValue(t) | ValueData::TimestampMicrosecondValue(t) => {
            ValueRef::Timestamp(Timestamp::new_microsecond(*t))
        }
        ValueData::TimestampNanosecondValue(t) => {
@@ -651,7 +646,6 @@ pub fn pb_values_to_vector_ref(data_type: &ConcreteDataType, values: Values) ->
        ConcreteDataType::Binary(_) => Arc::new(BinaryVector::from(values.binary_values)),
        ConcreteDataType::String(_) => Arc::new(StringVector::from_vec(values.string_values)),
        ConcreteDataType::Date(_) => Arc::new(DateVector::from_vec(values.date_values)),
-        ConcreteDataType::DateTime(_) => Arc::new(DateTimeVector::from_vec(values.datetime_values)),
        ConcreteDataType::Timestamp(unit) => match unit {
            TimestampType::Second(_) => Arc::new(TimestampSecondVector::from_vec(
                values.timestamp_second_values,
@@ -787,11 +781,6 @@ pub fn pb_values_to_values(data_type: &ConcreteDataType, values: Values) -> Vec<
            .into_iter()
            .map(|val| val.into())
            .collect(),
-        ConcreteDataType::DateTime(_) => values
-            .datetime_values
-            .into_iter()
-            .map(|v| Value::DateTime(v.into()))
-            .collect(),
        ConcreteDataType::Date(_) => values
            .date_values
            .into_iter()
@@ -947,9 +936,6 @@ pub fn to_proto_value(value: Value) -> Option<v1::Value> {
        Value::Date(v) => v1::Value {
            value_data: Some(ValueData::DateValue(v.val())),
        },
-        Value::DateTime(v) => v1::Value {
-            value_data: Some(ValueData::DatetimeValue(v.val())),
-        },
        Value::Timestamp(v) => match v.unit() {
            TimeUnit::Second => v1::Value {
                value_data: Some(ValueData::TimestampSecondValue(v.value())),
@@ -1066,7 +1052,6 @@ pub fn value_to_grpc_value(value: Value) -> GrpcValue {
            Value::String(v) => Some(ValueData::StringValue(v.as_utf8().to_string())),
            Value::Binary(v) => Some(ValueData::BinaryValue(v.to_vec())),
            Value::Date(v) => Some(ValueData::DateValue(v.val())),
-            Value::DateTime(v) => Some(ValueData::DatetimeValue(v.val())),
            Value::Timestamp(v) => Some(match v.unit() {
                TimeUnit::Second => ValueData::TimestampSecondValue(v.value()),
                TimeUnit::Millisecond => ValueData::TimestampMillisecondValue(v.value()),
@@ -1248,7 +1233,7 @@ mod tests {
            ColumnDataTypeWrapper::date_datatype().into()
        );
        assert_eq!(
-            ConcreteDataType::datetime_datatype(),
+            ConcreteDataType::timestamp_microsecond_datatype(),
            ColumnDataTypeWrapper::datetime_datatype().into()
        );
        assert_eq!(
@@ -1339,10 +1324,6 @@ mod tests {
            ColumnDataTypeWrapper::date_datatype(),
            ConcreteDataType::date_datatype().try_into().unwrap()
        );
-        assert_eq!(
-            ColumnDataTypeWrapper::datetime_datatype(),
-            ConcreteDataType::datetime_datatype().try_into().unwrap()
-        );
        assert_eq!(
            ColumnDataTypeWrapper::timestamp_millisecond_datatype(),
            ConcreteDataType::timestamp_millisecond_datatype()
@@ -1830,17 +1811,6 @@ mod tests {
        ]
    );

-    test_convert_values!(
-        datetime,
-        vec![1.into(), 2.into(), 3.into()],
-        datetime,
-        vec![
-            Value::DateTime(1.into()),
-            Value::DateTime(2.into()),
-            Value::DateTime(3.into())
-        ]
-    );
-
    #[test]
    fn test_vectors_to_rows_for_different_types() {
        let boolean_vec = BooleanVector::from_vec(vec![true, false, true]);
--- a/src/api/src/v1/column_def.rs
+++ b/src/api/src/v1/column_def.rs
@@ -132,6 +132,15 @@ pub fn options_from_skipping(skipping: &SkippingIndexOptions) -> Result<Option<C
    Ok((!options.options.is_empty()).then_some(options))
 }

+/// Tries to construct a `ColumnOptions` for inverted index.
+pub fn options_from_inverted() -> ColumnOptions {
+    let mut options = ColumnOptions::default();
+    options
+        .options
+        .insert(INVERTED_INDEX_GRPC_KEY.to_string(), "true".to_string());
+    options
+}
+
 /// Tries to construct a `FulltextAnalyzer` from the given analyzer.
 pub fn as_fulltext_option(analyzer: Analyzer) -> FulltextAnalyzer {
    match analyzer {
--- a/src/catalog/src/system_schema.rs
+++ b/src/catalog/src/system_schema.rs
@@ -77,7 +77,7 @@ trait SystemSchemaProviderInner {
    fn system_table(&self, name: &str) -> Option<SystemTableRef>;

    fn table_info(catalog_name: String, table: &SystemTableRef) -> TableInfoRef {
-        let table_meta = TableMetaBuilder::default()
+        let table_meta = TableMetaBuilder::empty()
            .schema(table.schema())
            .primary_key_indices(vec![])
            .next_column_id(0)
--- a/src/catalog/src/system_schema/information_schema/columns.rs
+++ b/src/catalog/src/system_schema/information_schema/columns.rs
@@ -365,10 +365,6 @@ impl InformationSchemaColumnsBuilder {
            self.numeric_scales.push(None);

            match &column_schema.data_type {
-                ConcreteDataType::DateTime(datetime_type) => {
-                    self.datetime_precisions
-                        .push(Some(datetime_type.precision() as i64));
-                }
                ConcreteDataType::Timestamp(ts_type) => {
                    self.datetime_precisions
                        .push(Some(ts_type.precision() as i64));
--- a/src/catalog/src/system_schema/information_schema/flows.rs
+++ b/src/catalog/src/system_schema/information_schema/flows.rs
@@ -28,16 +28,19 @@ use datafusion::physical_plan::streaming::PartitionStream as DfPartitionStream;
 use datatypes::prelude::ConcreteDataType as CDT;
 use datatypes::scalars::ScalarVectorBuilder;
 use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
+use datatypes::timestamp::TimestampMillisecond;
 use datatypes::value::Value;
 use datatypes::vectors::{
-    Int64VectorBuilder, StringVectorBuilder, UInt32VectorBuilder, UInt64VectorBuilder, VectorRef,
+    Int64VectorBuilder, StringVectorBuilder, TimestampMillisecondVectorBuilder,
+    UInt32VectorBuilder, UInt64VectorBuilder, VectorRef,
 };
 use futures::TryStreamExt;
 use snafu::{OptionExt, ResultExt};
 use store_api::storage::{ScanRequest, TableId};

 use crate::error::{
-    CreateRecordBatchSnafu, FlowInfoNotFoundSnafu, InternalSnafu, JsonSnafu, ListFlowsSnafu, Result,
+    CreateRecordBatchSnafu, FlowInfoNotFoundSnafu, InternalSnafu, JsonSnafu, ListFlowsSnafu,
+    Result, UpgradeWeakCatalogManagerRefSnafu,
 };
 use crate::information_schema::{Predicates, FLOWS};
 use crate::system_schema::information_schema::InformationTable;
@@ -59,6 +62,10 @@ pub const SOURCE_TABLE_IDS: &str = "source_table_ids";
 pub const SINK_TABLE_NAME: &str = "sink_table_name";
 pub const FLOWNODE_IDS: &str = "flownode_ids";
 pub const OPTIONS: &str = "options";
+pub const CREATED_TIME: &str = "created_time";
+pub const UPDATED_TIME: &str = "updated_time";
+pub const LAST_EXECUTION_TIME: &str = "last_execution_time";
+pub const SOURCE_TABLE_NAMES: &str = "source_table_names";

 /// The `information_schema.flows` to provides information about flows in databases.
 #[derive(Debug)]
@@ -99,6 +106,14 @@ impl InformationSchemaFlows {
                (SINK_TABLE_NAME, CDT::string_datatype(), false),
                (FLOWNODE_IDS, CDT::string_datatype(), true),
                (OPTIONS, CDT::string_datatype(), true),
+                (CREATED_TIME, CDT::timestamp_millisecond_datatype(), false),
+                (UPDATED_TIME, CDT::timestamp_millisecond_datatype(), false),
+                (
+                    LAST_EXECUTION_TIME,
+                    CDT::timestamp_millisecond_datatype(),
+                    true,
+                ),
+                (SOURCE_TABLE_NAMES, CDT::string_datatype(), true),
            ]
            .into_iter()
            .map(|(name, ty, nullable)| ColumnSchema::new(name, ty, nullable))
@@ -170,6 +185,10 @@ struct InformationSchemaFlowsBuilder {
    sink_table_names: StringVectorBuilder,
    flownode_id_groups: StringVectorBuilder,
    option_groups: StringVectorBuilder,
+    created_time: TimestampMillisecondVectorBuilder,
+    updated_time: TimestampMillisecondVectorBuilder,
+    last_execution_time: TimestampMillisecondVectorBuilder,
+    source_table_names: StringVectorBuilder,
 }

 impl InformationSchemaFlowsBuilder {
@@ -196,6 +215,10 @@ impl InformationSchemaFlowsBuilder {
            sink_table_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
            flownode_id_groups: StringVectorBuilder::with_capacity(INIT_CAPACITY),
            option_groups: StringVectorBuilder::with_capacity(INIT_CAPACITY),
+            created_time: TimestampMillisecondVectorBuilder::with_capacity(INIT_CAPACITY),
+            updated_time: TimestampMillisecondVectorBuilder::with_capacity(INIT_CAPACITY),
+            last_execution_time: TimestampMillisecondVectorBuilder::with_capacity(INIT_CAPACITY),
+            source_table_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
        }
    }

@@ -235,13 +258,14 @@ impl InformationSchemaFlowsBuilder {
                    catalog_name: catalog_name.to_string(),
                    flow_name: flow_name.to_string(),
                })?;
-            self.add_flow(&predicates, flow_id.flow_id(), flow_info, &flow_stat)?;
+            self.add_flow(&predicates, flow_id.flow_id(), flow_info, &flow_stat)
+                .await?;
        }

        self.finish()
    }

-    fn add_flow(
+    async fn add_flow(
        &mut self,
        predicates: &Predicates,
        flow_id: FlowId,
@@ -290,6 +314,36 @@ impl InformationSchemaFlowsBuilder {
                    input: format!("{:?}", flow_info.options()),
                },
            )?));
+        self.created_time
+            .push(Some(flow_info.created_time().timestamp_millis().into()));
+        self.updated_time
+            .push(Some(flow_info.updated_time().timestamp_millis().into()));
+        self.last_execution_time
+            .push(flow_stat.as_ref().and_then(|state| {
+                state
+                    .last_exec_time_map
+                    .get(&flow_id)
+                    .map(|v| TimestampMillisecond::new(*v))
+            }));
+
+        let mut source_table_names = vec![];
+        let catalog_name = self.catalog_name.clone();
+        let catalog_manager = self
+            .catalog_manager
+            .upgrade()
+            .context(UpgradeWeakCatalogManagerRefSnafu)?;
+        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
+            source_table_names.extend(
+                catalog_manager
+                    .tables_by_ids(&catalog_name, &schema_name, flow_info.source_table_ids())
+                    .await?
+                    .into_iter()
+                    .map(|table| table.table_info().full_table_name()),
+            );
+        }
+
+        let source_table_names = source_table_names.join(",");
+        self.source_table_names.push(Some(&source_table_names));

        Ok(())
    }
@@ -307,6 +361,10 @@ impl InformationSchemaFlowsBuilder {
            Arc::new(self.sink_table_names.finish()),
            Arc::new(self.flownode_id_groups.finish()),
            Arc::new(self.option_groups.finish()),
+            Arc::new(self.created_time.finish()),
+            Arc::new(self.updated_time.finish()),
+            Arc::new(self.last_execution_time.finish()),
+            Arc::new(self.source_table_names.finish()),
        ];
        RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
    }
--- a/src/catalog/src/system_schema/information_schema/information_memory_table.rs
+++ b/src/catalog/src/system_schema/information_schema/information_memory_table.rs
@@ -20,7 +20,7 @@ use datatypes::vectors::{Int64Vector, StringVector, VectorRef};

 use super::table_names::*;
 use crate::system_schema::utils::tables::{
-    bigint_column, datetime_column, string_column, string_columns,
+    bigint_column, string_column, string_columns, timestamp_micro_column,
 };

 const NO_VALUE: &str = "NO";
@@ -163,17 +163,17 @@ pub(super) fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>
                string_column("EVENT_BODY"),
                string_column("EVENT_DEFINITION"),
                string_column("EVENT_TYPE"),
-                datetime_column("EXECUTE_AT"),
+                timestamp_micro_column("EXECUTE_AT"),
                bigint_column("INTERVAL_VALUE"),
                string_column("INTERVAL_FIELD"),
                string_column("SQL_MODE"),
-                datetime_column("STARTS"),
-                datetime_column("ENDS"),
+                timestamp_micro_column("STARTS"),
+                timestamp_micro_column("ENDS"),
                string_column("STATUS"),
                string_column("ON_COMPLETION"),
-                datetime_column("CREATED"),
-                datetime_column("LAST_ALTERED"),
-                datetime_column("LAST_EXECUTED"),
+                timestamp_micro_column("CREATED"),
+                timestamp_micro_column("LAST_ALTERED"),
+                timestamp_micro_column("LAST_EXECUTED"),
                string_column("EVENT_COMMENT"),
                bigint_column("ORIGINATOR"),
                string_column("CHARACTER_SET_CLIENT"),
@@ -204,10 +204,10 @@ pub(super) fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>
                bigint_column("INITIAL_SIZE"),
                bigint_column("MAXIMUM_SIZE"),
                bigint_column("AUTOEXTEND_SIZE"),
-                datetime_column("CREATION_TIME"),
-                datetime_column("LAST_UPDATE_TIME"),
-                datetime_column("LAST_ACCESS_TIME"),
-                datetime_column("RECOVER_TIME"),
+                timestamp_micro_column("CREATION_TIME"),
+                timestamp_micro_column("LAST_UPDATE_TIME"),
+                timestamp_micro_column("LAST_ACCESS_TIME"),
+                timestamp_micro_column("RECOVER_TIME"),
                bigint_column("TRANSACTION_COUNTER"),
                string_column("VERSION"),
                string_column("ROW_FORMAT"),
@@ -217,9 +217,9 @@ pub(super) fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>
                bigint_column("MAX_DATA_LENGTH"),
                bigint_column("INDEX_LENGTH"),
                bigint_column("DATA_FREE"),
-                datetime_column("CREATE_TIME"),
-                datetime_column("UPDATE_TIME"),
-                datetime_column("CHECK_TIME"),
+                timestamp_micro_column("CREATE_TIME"),
+                timestamp_micro_column("UPDATE_TIME"),
+                timestamp_micro_column("CHECK_TIME"),
                string_column("CHECKSUM"),
                string_column("STATUS"),
                string_column("EXTRA"),
@@ -330,8 +330,8 @@ pub(super) fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>
                string_column("SQL_DATA_ACCESS"),
                string_column("SQL_PATH"),
                string_column("SECURITY_TYPE"),
-                datetime_column("CREATED"),
-                datetime_column("LAST_ALTERED"),
+                timestamp_micro_column("CREATED"),
+                timestamp_micro_column("LAST_ALTERED"),
                string_column("SQL_MODE"),
                string_column("ROUTINE_COMMENT"),
                string_column("DEFINER"),
@@ -383,7 +383,7 @@ pub(super) fn get_schema_columns(table_name: &str) -> (SchemaRef, Vec<VectorRef>
                string_column("ACTION_REFERENCE_NEW_TABLE"),
                string_column("ACTION_REFERENCE_OLD_ROW"),
                string_column("ACTION_REFERENCE_NEW_ROW"),
-                datetime_column("CREATED"),
+                timestamp_micro_column("CREATED"),
                string_column("SQL_MODE"),
                string_column("DEFINER"),
                string_column("CHARACTER_SET_CLIENT"),
--- a/src/catalog/src/system_schema/information_schema/partitions.rs
+++ b/src/catalog/src/system_schema/information_schema/partitions.rs
@@ -20,17 +20,18 @@ use common_catalog::consts::INFORMATION_SCHEMA_PARTITIONS_TABLE_ID;
 use common_error::ext::BoxedError;
 use common_recordbatch::adapter::RecordBatchStreamAdapter;
 use common_recordbatch::{RecordBatch, SendableRecordBatchStream};
-use common_time::datetime::DateTime;
 use datafusion::execution::TaskContext;
 use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
 use datafusion::physical_plan::streaming::PartitionStream as DfPartitionStream;
 use datafusion::physical_plan::SendableRecordBatchStream as DfSendableRecordBatchStream;
 use datatypes::prelude::{ConcreteDataType, ScalarVectorBuilder, VectorRef};
 use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
+use datatypes::timestamp::TimestampMicrosecond;
 use datatypes::value::Value;
 use datatypes::vectors::{
-    ConstantVector, DateTimeVector, DateTimeVectorBuilder, Int64Vector, Int64VectorBuilder,
-    MutableVector, StringVector, StringVectorBuilder, UInt64VectorBuilder,
+    ConstantVector, Int64Vector, Int64VectorBuilder, MutableVector, StringVector,
+    StringVectorBuilder, TimestampMicrosecondVector, TimestampMicrosecondVectorBuilder,
+    UInt64VectorBuilder,
 };
 use futures::{StreamExt, TryStreamExt};
 use partition::manager::PartitionInfo;
@@ -127,9 +128,21 @@ impl InformationSchemaPartitions {
            ColumnSchema::new("max_data_length", ConcreteDataType::int64_datatype(), true),
            ColumnSchema::new("index_length", ConcreteDataType::int64_datatype(), true),
            ColumnSchema::new("data_free", ConcreteDataType::int64_datatype(), true),
-            ColumnSchema::new("create_time", ConcreteDataType::datetime_datatype(), true),
-            ColumnSchema::new("update_time", ConcreteDataType::datetime_datatype(), true),
-            ColumnSchema::new("check_time", ConcreteDataType::datetime_datatype(), true),
+            ColumnSchema::new(
+                "create_time",
+                ConcreteDataType::timestamp_microsecond_datatype(),
+                true,
+            ),
+            ColumnSchema::new(
+                "update_time",
+                ConcreteDataType::timestamp_microsecond_datatype(),
+                true,
+            ),
+            ColumnSchema::new(
+                "check_time",
+                ConcreteDataType::timestamp_microsecond_datatype(),
+                true,
+            ),
            ColumnSchema::new("checksum", ConcreteDataType::int64_datatype(), true),
            ColumnSchema::new(
                "partition_comment",
@@ -200,7 +213,7 @@ struct InformationSchemaPartitionsBuilder {
    partition_names: StringVectorBuilder,
    partition_ordinal_positions: Int64VectorBuilder,
    partition_expressions: StringVectorBuilder,
-    create_times: DateTimeVectorBuilder,
+    create_times: TimestampMicrosecondVectorBuilder,
    partition_ids: UInt64VectorBuilder,
 }

@@ -220,7 +233,7 @@ impl InformationSchemaPartitionsBuilder {
            partition_names: StringVectorBuilder::with_capacity(INIT_CAPACITY),
            partition_ordinal_positions: Int64VectorBuilder::with_capacity(INIT_CAPACITY),
            partition_expressions: StringVectorBuilder::with_capacity(INIT_CAPACITY),
-            create_times: DateTimeVectorBuilder::with_capacity(INIT_CAPACITY),
+            create_times: TimestampMicrosecondVectorBuilder::with_capacity(INIT_CAPACITY),
            partition_ids: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
        }
    }
@@ -324,7 +337,7 @@ impl InformationSchemaPartitionsBuilder {
            };

            self.partition_expressions.push(expressions.as_deref());
-            self.create_times.push(Some(DateTime::from(
+            self.create_times.push(Some(TimestampMicrosecond::from(
                table_info.meta.created_on.timestamp_millis(),
            )));
            self.partition_ids.push(Some(partition.id.as_u64()));
@@ -342,8 +355,8 @@ impl InformationSchemaPartitionsBuilder {
            Arc::new(Int64Vector::from(vec![None])),
            rows_num,
        ));
-        let null_datetime_vector = Arc::new(ConstantVector::new(
-            Arc::new(DateTimeVector::from(vec![None])),
+        let null_timestampmicrosecond_vector = Arc::new(ConstantVector::new(
+            Arc::new(TimestampMicrosecondVector::from(vec![None])),
            rows_num,
        ));
        let partition_methods = Arc::new(ConstantVector::new(
@@ -373,8 +386,8 @@ impl InformationSchemaPartitionsBuilder {
            null_i64_vector.clone(),
            Arc::new(self.create_times.finish()),
            // TODO(dennis): supports update_time
-            null_datetime_vector.clone(),
-            null_datetime_vector,
+            null_timestampmicrosecond_vector.clone(),
+            null_timestampmicrosecond_vector,
            null_i64_vector,
            null_string_vector.clone(),
            null_string_vector.clone(),
--- a/src/catalog/src/system_schema/information_schema/tables.rs
+++ b/src/catalog/src/system_schema/information_schema/tables.rs
@@ -30,7 +30,8 @@ use datatypes::prelude::{ConcreteDataType, ScalarVectorBuilder, VectorRef};
 use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
 use datatypes::value::Value;
 use datatypes::vectors::{
-    DateTimeVectorBuilder, StringVectorBuilder, UInt32VectorBuilder, UInt64VectorBuilder,
+    StringVectorBuilder, TimestampMicrosecondVectorBuilder, UInt32VectorBuilder,
+    UInt64VectorBuilder,
 };
 use futures::TryStreamExt;
 use snafu::{OptionExt, ResultExt};
@@ -105,9 +106,21 @@ impl InformationSchemaTables {
            ColumnSchema::new(TABLE_ROWS, ConcreteDataType::uint64_datatype(), true),
            ColumnSchema::new(DATA_FREE, ConcreteDataType::uint64_datatype(), true),
            ColumnSchema::new(AUTO_INCREMENT, ConcreteDataType::uint64_datatype(), true),
-            ColumnSchema::new(CREATE_TIME, ConcreteDataType::datetime_datatype(), true),
-            ColumnSchema::new(UPDATE_TIME, ConcreteDataType::datetime_datatype(), true),
-            ColumnSchema::new(CHECK_TIME, ConcreteDataType::datetime_datatype(), true),
+            ColumnSchema::new(
+                CREATE_TIME,
+                ConcreteDataType::timestamp_microsecond_datatype(),
+                true,
+            ),
+            ColumnSchema::new(
+                UPDATE_TIME,
+                ConcreteDataType::timestamp_microsecond_datatype(),
+                true,
+            ),
+            ColumnSchema::new(
+                CHECK_TIME,
+                ConcreteDataType::timestamp_microsecond_datatype(),
+                true,
+            ),
            ColumnSchema::new(TABLE_COLLATION, ConcreteDataType::string_datatype(), true),
            ColumnSchema::new(CHECKSUM, ConcreteDataType::uint64_datatype(), true),
            ColumnSchema::new(CREATE_OPTIONS, ConcreteDataType::string_datatype(), true),
@@ -182,9 +195,9 @@ struct InformationSchemaTablesBuilder {
    max_index_length: UInt64VectorBuilder,
    data_free: UInt64VectorBuilder,
    auto_increment: UInt64VectorBuilder,
-    create_time: DateTimeVectorBuilder,
-    update_time: DateTimeVectorBuilder,
-    check_time: DateTimeVectorBuilder,
+    create_time: TimestampMicrosecondVectorBuilder,
+    update_time: TimestampMicrosecondVectorBuilder,
+    check_time: TimestampMicrosecondVectorBuilder,
    table_collation: StringVectorBuilder,
    checksum: UInt64VectorBuilder,
    create_options: StringVectorBuilder,
@@ -219,9 +232,9 @@ impl InformationSchemaTablesBuilder {
            max_index_length: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
            data_free: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
            auto_increment: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
-            create_time: DateTimeVectorBuilder::with_capacity(INIT_CAPACITY),
-            update_time: DateTimeVectorBuilder::with_capacity(INIT_CAPACITY),
-            check_time: DateTimeVectorBuilder::with_capacity(INIT_CAPACITY),
+            create_time: TimestampMicrosecondVectorBuilder::with_capacity(INIT_CAPACITY),
+            update_time: TimestampMicrosecondVectorBuilder::with_capacity(INIT_CAPACITY),
+            check_time: TimestampMicrosecondVectorBuilder::with_capacity(INIT_CAPACITY),
            table_collation: StringVectorBuilder::with_capacity(INIT_CAPACITY),
            checksum: UInt64VectorBuilder::with_capacity(INIT_CAPACITY),
            create_options: StringVectorBuilder::with_capacity(INIT_CAPACITY),
--- a/src/catalog/src/system_schema/utils/tables.rs
+++ b/src/catalog/src/system_schema/utils/tables.rs
@@ -51,10 +51,10 @@ pub fn bigint_column(name: &str) -> ColumnSchema {
    )
 }

-pub fn datetime_column(name: &str) -> ColumnSchema {
+pub fn timestamp_micro_column(name: &str) -> ColumnSchema {
    ColumnSchema::new(
        str::to_lowercase(name),
-        ConcreteDataType::datetime_datatype(),
+        ConcreteDataType::timestamp_microsecond_datatype(),
        false,
    )
 }
--- a/src/cli/Cargo.toml
+++ b/src/cli/Cargo.toml
@@ -6,6 +6,7 @@ license.workspace = true

 [features]
 pg_kvbackend = ["common-meta/pg_kvbackend"]
+mysql_kvbackend = ["common-meta/mysql_kvbackend"]

 [lints]
 workspace = true
@@ -43,6 +44,10 @@ futures.workspace = true
 humantime.workspace = true
 meta-client.workspace = true
 nu-ansi-term = "0.46"
+opendal = { version = "0.51.1", features = [
+    "services-fs",
+    "services-s3",
+] }
 query.workspace = true
 rand.workspace = true
 reqwest.workspace = true
--- a/src/cli/src/bench.rs
+++ b/src/cli/src/bench.rs
@@ -23,6 +23,8 @@ use common_error::ext::BoxedError;
 use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
 use common_meta::kv_backend::etcd::EtcdStore;
 use common_meta::kv_backend::memory::MemoryKvBackend;
+#[cfg(feature = "mysql_kvbackend")]
+use common_meta::kv_backend::rds::MySqlStore;
 #[cfg(feature = "pg_kvbackend")]
 use common_meta::kv_backend::rds::PgStore;
 use common_meta::peer::Peer;
@@ -63,6 +65,9 @@ pub struct BenchTableMetadataCommand {
    #[cfg(feature = "pg_kvbackend")]
    #[clap(long)]
    postgres_addr: Option<String>,
+    #[cfg(feature = "mysql_kvbackend")]
+    #[clap(long)]
+    mysql_addr: Option<String>,
    #[clap(long)]
    count: u32,
 }
@@ -86,6 +91,16 @@ impl BenchTableMetadataCommand {
            kv_backend
        };

+        #[cfg(feature = "mysql_kvbackend")]
+        let kv_backend = if let Some(mysql_addr) = &self.mysql_addr {
+            info!("Using mysql as kv backend");
+            MySqlStore::with_url(mysql_addr, "greptime_metakv", 128)
+                .await
+                .unwrap()
+        } else {
+            kv_backend
+        };
+
        let table_metadata_manager = Arc::new(TableMetadataManager::new(kv_backend));

        let tool = BenchTableMetadata {
--- a/src/cli/src/error.rs
+++ b/src/cli/src/error.rs
@@ -276,6 +276,24 @@ pub enum Error {
        #[snafu(implicit)]
        location: Location,
    },
+
+    #[snafu(display("OpenDAL operator failed"))]
+    OpenDal {
+        #[snafu(implicit)]
+        location: Location,
+        #[snafu(source)]
+        error: opendal::Error,
+    },
+    #[snafu(display("S3 config need be set"))]
+    S3ConfigNotSet {
+        #[snafu(implicit)]
+        location: Location,
+    },
+    #[snafu(display("Output directory not set"))]
+    OutputDirNotSet {
+        #[snafu(implicit)]
+        location: Location,
+    },
 }

 pub type Result<T> = std::result::Result<T, Error>;
@@ -319,6 +337,9 @@ impl ErrorExt for Error {
            | Error::BuildClient { .. } => StatusCode::Unexpected,

            Error::Other { source, .. } => source.status_code(),
+            Error::OpenDal { .. } => StatusCode::Internal,
+            Error::S3ConfigNotSet { .. } => StatusCode::InvalidArguments,
+            Error::OutputDirNotSet { .. } => StatusCode::InvalidArguments,

            Error::BuildRuntime { source, .. } => source.status_code(),

--- a/src/cli/src/export.rs
+++ b/src/cli/src/export.rs
@@ -21,15 +21,18 @@ use async_trait::async_trait;
 use clap::{Parser, ValueEnum};
 use common_error::ext::BoxedError;
 use common_telemetry::{debug, error, info};
+use opendal::layers::LoggingLayer;
+use opendal::{services, Operator};
 use serde_json::Value;
 use snafu::{OptionExt, ResultExt};
-use tokio::fs::File;
-use tokio::io::{AsyncWriteExt, BufWriter};
 use tokio::sync::Semaphore;
 use tokio::time::Instant;

 use crate::database::{parse_proxy_opts, DatabaseClient};
-use crate::error::{EmptyResultSnafu, Error, FileIoSnafu, Result, SchemaNotFoundSnafu};
+use crate::error::{
+    EmptyResultSnafu, Error, OpenDalSnafu, OutputDirNotSetSnafu, Result, S3ConfigNotSetSnafu,
+    SchemaNotFoundSnafu,
+};
 use crate::{database, Tool};

 type TableReference = (String, String, String);
@@ -52,8 +55,9 @@ pub struct ExportCommand {
    addr: String,

    /// Directory to put the exported data. E.g.: /tmp/greptimedb-export
+    /// for local export.
    #[clap(long)]
-    output_dir: String,
+    output_dir: Option<String>,

    /// The name of the catalog to export.
    #[clap(long, default_value = "greptime-*")]
@@ -101,10 +105,51 @@ pub struct ExportCommand {
    /// Disable proxy server, if set, will not use any proxy.
    #[clap(long)]
    no_proxy: bool,
+
+    /// if export data to s3
+    #[clap(long)]
+    s3: bool,
+
+    /// The s3 bucket name
+    /// if s3 is set, this is required
+    #[clap(long)]
+    s3_bucket: Option<String>,
+
+    /// The s3 endpoint
+    /// if s3 is set, this is required
+    #[clap(long)]
+    s3_endpoint: Option<String>,
+
+    /// The s3 access key
+    /// if s3 is set, this is required
+    #[clap(long)]
+    s3_access_key: Option<String>,
+
+    /// The s3 secret key
+    /// if s3 is set, this is required
+    #[clap(long)]
+    s3_secret_key: Option<String>,
+
+    /// The s3 region
+    /// if s3 is set, this is required
+    #[clap(long)]
+    s3_region: Option<String>,
 }

 impl ExportCommand {
    pub async fn build(&self) -> std::result::Result<Box<dyn Tool>, BoxedError> {
+        if self.s3
+            && (self.s3_bucket.is_none()
+                || self.s3_endpoint.is_none()
+                || self.s3_access_key.is_none()
+                || self.s3_secret_key.is_none()
+                || self.s3_region.is_none())
+        {
+            return Err(BoxedError::new(S3ConfigNotSetSnafu {}.build()));
+        }
+        if !self.s3 && self.output_dir.is_none() {
+            return Err(BoxedError::new(OutputDirNotSetSnafu {}.build()));
+        }
        let (catalog, schema) =
            database::split_database(&self.database).map_err(BoxedError::new)?;
        let proxy = parse_proxy_opts(self.proxy.clone(), self.no_proxy)?;
@@ -126,24 +171,43 @@ impl ExportCommand {
            target: self.target.clone(),
            start_time: self.start_time.clone(),
            end_time: self.end_time.clone(),
+            s3: self.s3,
+            s3_bucket: self.s3_bucket.clone(),
+            s3_endpoint: self.s3_endpoint.clone(),
+            s3_access_key: self.s3_access_key.clone(),
+            s3_secret_key: self.s3_secret_key.clone(),
+            s3_region: self.s3_region.clone(),
        }))
    }
 }

+#[derive(Clone)]
 pub struct Export {
    catalog: String,
    schema: Option<String>,
    database_client: DatabaseClient,
-    output_dir: String,
+    output_dir: Option<String>,
    parallelism: usize,
    target: ExportTarget,
    start_time: Option<String>,
    end_time: Option<String>,
+    s3: bool,
+    s3_bucket: Option<String>,
+    s3_endpoint: Option<String>,
+    s3_access_key: Option<String>,
+    s3_secret_key: Option<String>,
+    s3_region: Option<String>,
 }

 impl Export {
    fn catalog_path(&self) -> PathBuf {
-        PathBuf::from(&self.output_dir).join(&self.catalog)
+        if self.s3 {
+            PathBuf::from(&self.catalog)
+        } else if let Some(dir) = &self.output_dir {
+            PathBuf::from(dir).join(&self.catalog)
+        } else {
+            unreachable!("catalog_path: output_dir must be set when not using s3")
+        }
    }

    async fn get_db_names(&self) -> Result<Vec<String>> {
@@ -300,19 +364,23 @@ impl Export {
        let timer = Instant::now();
        let db_names = self.get_db_names().await?;
        let db_count = db_names.len();
+        let operator = self.build_operator().await?;
+
        for schema in db_names {
-            let db_dir = self.catalog_path().join(format!("{schema}/"));
-            tokio::fs::create_dir_all(&db_dir)
-                .await
-                .context(FileIoSnafu)?;
-            let file = db_dir.join("create_database.sql");
-            let mut file = File::create(file).await.context(FileIoSnafu)?;
            let create_database = self
                .show_create("DATABASE", &self.catalog, &schema, None)
                .await?;
-            file.write_all(create_database.as_bytes())
-                .await
-                .context(FileIoSnafu)?;
+
+            let file_path = self.get_file_path(&schema, "create_database.sql");
+            self.write_to_storage(&operator, &file_path, create_database.into_bytes())
+                .await?;
+
+            info!(
+                "Exported {}.{} database creation SQL to {}",
+                self.catalog,
+                schema,
+                self.format_output_path(&file_path)
+            );
        }

        let elapsed = timer.elapsed();
@@ -326,149 +394,267 @@ impl Export {
        let semaphore = Arc::new(Semaphore::new(self.parallelism));
        let db_names = self.get_db_names().await?;
        let db_count = db_names.len();
+        let operator = Arc::new(self.build_operator().await?);
        let mut tasks = Vec::with_capacity(db_names.len());
+
        for schema in db_names {
            let semaphore_moved = semaphore.clone();
+            let export_self = self.clone();
+            let operator = operator.clone();
            tasks.push(async move {
                let _permit = semaphore_moved.acquire().await.unwrap();
-                let (metric_physical_tables, remaining_tables, views) =
-                    self.get_table_list(&self.catalog, &schema).await?;
-                let table_count =
-                    metric_physical_tables.len() + remaining_tables.len() + views.len();
-                let db_dir = self.catalog_path().join(format!("{schema}/"));
-                tokio::fs::create_dir_all(&db_dir)
-                    .await
-                    .context(FileIoSnafu)?;
-                let file = db_dir.join("create_tables.sql");
-                let mut file = File::create(file).await.context(FileIoSnafu)?;
-                for (c, s, t) in metric_physical_tables.into_iter().chain(remaining_tables) {
-                    let create_table = self.show_create("TABLE", &c, &s, Some(&t)).await?;
-                    file.write_all(create_table.as_bytes())
-                        .await
-                        .context(FileIoSnafu)?;
-                }
-                for (c, s, v) in views {
-                    let create_view = self.show_create("VIEW", &c, &s, Some(&v)).await?;
-                    file.write_all(create_view.as_bytes())
-                        .await
-                        .context(FileIoSnafu)?;
+                let (metric_physical_tables, remaining_tables, views) = export_self
+                    .get_table_list(&export_self.catalog, &schema)
+                    .await?;
+
+                // Create directory if needed for file system storage
+                if !export_self.s3 {
+                    let db_dir = format!("{}/{}/", export_self.catalog, schema);
+                    operator.create_dir(&db_dir).await.context(OpenDalSnafu)?;
                }

+                let file_path = export_self.get_file_path(&schema, "create_tables.sql");
+                let mut content = Vec::new();
+
+                // Add table creation SQL
+                for (c, s, t) in metric_physical_tables.iter().chain(&remaining_tables) {
+                    let create_table = export_self.show_create("TABLE", c, s, Some(t)).await?;
+                    content.extend_from_slice(create_table.as_bytes());
+                }
+
+                // Add view creation SQL
+                for (c, s, v) in &views {
+                    let create_view = export_self.show_create("VIEW", c, s, Some(v)).await?;
+                    content.extend_from_slice(create_view.as_bytes());
+                }
+
+                // Write to storage
+                export_self
+                    .write_to_storage(&operator, &file_path, content)
+                    .await?;
+
                info!(
-                    "Finished exporting {}.{schema} with {table_count} table schemas to path: {}",
-                    self.catalog,
-                    db_dir.to_string_lossy()
+                    "Finished exporting {}.{schema} with {} table schemas to path: {}",
+                    export_self.catalog,
+                    metric_physical_tables.len() + remaining_tables.len() + views.len(),
+                    export_self.format_output_path(&file_path)
                );

                Ok::<(), Error>(())
            });
        }

-        let success = futures::future::join_all(tasks)
-            .await
-            .into_iter()
-            .filter(|r| match r {
-                Ok(_) => true,
-                Err(e) => {
-                    error!(e; "export schema job failed");
-                    false
-                }
-            })
-            .count();
-
+        let success = self.execute_tasks(tasks).await;
        let elapsed = timer.elapsed();
        info!("Success {success}/{db_count} jobs, cost: {elapsed:?}");

        Ok(())
    }

+    async fn build_operator(&self) -> Result<Operator> {
+        if self.s3 {
+            self.build_s3_operator().await
+        } else {
+            self.build_fs_operator().await
+        }
+    }
+
+    async fn build_s3_operator(&self) -> Result<Operator> {
+        let mut builder = services::S3::default().root("").bucket(
+            self.s3_bucket
+                .as_ref()
+                .expect("s3_bucket must be provided when s3 is enabled"),
+        );
+
+        if let Some(endpoint) = self.s3_endpoint.as_ref() {
+            builder = builder.endpoint(endpoint);
+        }
+
+        if let Some(region) = self.s3_region.as_ref() {
+            builder = builder.region(region);
+        }
+
+        if let Some(key_id) = self.s3_access_key.as_ref() {
+            builder = builder.access_key_id(key_id);
+        }
+
+        if let Some(secret_key) = self.s3_secret_key.as_ref() {
+            builder = builder.secret_access_key(secret_key);
+        }
+
+        let op = Operator::new(builder)
+            .context(OpenDalSnafu)?
+            .layer(LoggingLayer::default())
+            .finish();
+        Ok(op)
+    }
+
+    async fn build_fs_operator(&self) -> Result<Operator> {
+        let root = self
+            .output_dir
+            .as_ref()
+            .context(OutputDirNotSetSnafu)?
+            .clone();
+        let op = Operator::new(services::Fs::default().root(&root))
+            .context(OpenDalSnafu)?
+            .layer(LoggingLayer::default())
+            .finish();
+        Ok(op)
+    }
+
    async fn export_database_data(&self) -> Result<()> {
        let timer = Instant::now();
        let semaphore = Arc::new(Semaphore::new(self.parallelism));
        let db_names = self.get_db_names().await?;
        let db_count = db_names.len();
        let mut tasks = Vec::with_capacity(db_count);
+        let operator = Arc::new(self.build_operator().await?);
+        let with_options = build_with_options(&self.start_time, &self.end_time);
+
        for schema in db_names {
            let semaphore_moved = semaphore.clone();
+            let export_self = self.clone();
+            let with_options_clone = with_options.clone();
+            let operator = operator.clone();
+
            tasks.push(async move {
                let _permit = semaphore_moved.acquire().await.unwrap();
-                let db_dir = self.catalog_path().join(format!("{schema}/"));
-                tokio::fs::create_dir_all(&db_dir)
-                    .await
-                    .context(FileIoSnafu)?;

-                let with_options = match (&self.start_time, &self.end_time) {
-                    (Some(start_time), Some(end_time)) => {
-                        format!(
-                            "WITH (FORMAT='parquet', start_time='{}', end_time='{}')",
-                            start_time, end_time
-                        )
-                    }
-                    (Some(start_time), None) => {
-                        format!("WITH (FORMAT='parquet', start_time='{}')", start_time)
-                    }
-                    (None, Some(end_time)) => {
-                        format!("WITH (FORMAT='parquet', end_time='{}')", end_time)
-                    }
-                    (None, None) => "WITH (FORMAT='parquet')".to_string(),
-                };
+                // Create directory if not using S3
+                if !export_self.s3 {
+                    let db_dir = format!("{}/{}/", export_self.catalog, schema);
+                    operator.create_dir(&db_dir).await.context(OpenDalSnafu)?;
+                }

+                let (path, connection_part) = export_self.get_storage_params(&schema);
+
+                // Execute COPY DATABASE TO command
                let sql = format!(
-                    r#"COPY DATABASE "{}"."{}" TO '{}' {};"#,
-                    self.catalog,
-                    schema,
-                    db_dir.to_str().unwrap(),
-                    with_options
+                    r#"COPY DATABASE "{}"."{}" TO '{}' WITH ({}){};"#,
+                    export_self.catalog, schema, path, with_options_clone, connection_part
+                );
+                info!("Executing sql: {sql}");
+                export_self.database_client.sql_in_public(&sql).await?;
+                info!(
+                    "Finished exporting {}.{} data to {}",
+                    export_self.catalog, schema, path
                );

-                info!("Executing sql: {sql}");
+                // Create copy_from.sql file
+                let copy_database_from_sql = format!(
+                    r#"COPY DATABASE "{}"."{}" FROM '{}' WITH ({}){};"#,
+                    export_self.catalog, schema, path, with_options_clone, connection_part
+                );

-                self.database_client.sql_in_public(&sql).await?;
+                let copy_from_path = export_self.get_file_path(&schema, "copy_from.sql");
+                export_self
+                    .write_to_storage(
+                        &operator,
+                        &copy_from_path,
+                        copy_database_from_sql.into_bytes(),
+                    )
+                    .await?;

                info!(
-                    "Finished exporting {}.{schema} data into path: {}",
-                    self.catalog,
-                    db_dir.to_string_lossy()
-                );
-
-                // The export copy from sql
-                let copy_from_file = db_dir.join("copy_from.sql");
-                let mut writer =
-                    BufWriter::new(File::create(copy_from_file).await.context(FileIoSnafu)?);
-                let copy_database_from_sql = format!(
-                    r#"COPY DATABASE "{}"."{}" FROM '{}' WITH (FORMAT='parquet');"#,
-                    self.catalog,
+                    "Finished exporting {}.{} copy_from.sql to {}",
+                    export_self.catalog,
                    schema,
-                    db_dir.to_str().unwrap()
+                    export_self.format_output_path(&copy_from_path)
                );
-                writer
-                    .write(copy_database_from_sql.as_bytes())
-                    .await
-                    .context(FileIoSnafu)?;
-                writer.flush().await.context(FileIoSnafu)?;
-
-                info!("Finished exporting {}.{schema} copy_from.sql", self.catalog);

                Ok::<(), Error>(())
-            })
+            });
        }

-        let success = futures::future::join_all(tasks)
+        let success = self.execute_tasks(tasks).await;
+        let elapsed = timer.elapsed();
+        info!("Success {success}/{db_count} jobs, costs: {elapsed:?}");
+
+        Ok(())
+    }
+
+    fn get_file_path(&self, schema: &str, file_name: &str) -> String {
+        format!("{}/{}/{}", self.catalog, schema, file_name)
+    }
+
+    fn format_output_path(&self, file_path: &str) -> String {
+        if self.s3 {
+            format!(
+                "s3://{}/{}",
+                self.s3_bucket.as_ref().unwrap_or(&String::new()),
+                file_path
+            )
+        } else {
+            format!(
+                "{}/{}",
+                self.output_dir.as_ref().unwrap_or(&String::new()),
+                file_path
+            )
+        }
+    }
+
+    async fn write_to_storage(
+        &self,
+        op: &Operator,
+        file_path: &str,
+        content: Vec<u8>,
+    ) -> Result<()> {
+        op.write(file_path, content).await.context(OpenDalSnafu)
+    }
+
+    fn get_storage_params(&self, schema: &str) -> (String, String) {
+        if self.s3 {
+            let s3_path = format!(
+                "s3://{}/{}/{}/",
+                // Safety: s3_bucket is required when s3 is enabled
+                self.s3_bucket.as_ref().unwrap(),
+                self.catalog,
+                schema
+            );
+
+            // endpoint is optional
+            let endpoint_option = if let Some(endpoint) = self.s3_endpoint.as_ref() {
+                format!(", ENDPOINT='{}'", endpoint)
+            } else {
+                String::new()
+            };
+
+            // Safety: All s3 options are required
+            let connection_options = format!(
+                "ACCESS_KEY_ID='{}', SECRET_ACCESS_KEY='{}', REGION='{}'{}",
+                self.s3_access_key.as_ref().unwrap(),
+                self.s3_secret_key.as_ref().unwrap(),
+                self.s3_region.as_ref().unwrap(),
+                endpoint_option
+            );
+
+            (s3_path, format!(" CONNECTION ({})", connection_options))
+        } else {
+            (
+                self.catalog_path()
+                    .join(format!("{schema}/"))
+                    .to_string_lossy()
+                    .to_string(),
+                String::new(),
+            )
+        }
+    }
+
+    async fn execute_tasks(
+        &self,
+        tasks: Vec<impl std::future::Future<Output = Result<()>>>,
+    ) -> usize {
+        futures::future::join_all(tasks)
            .await
            .into_iter()
            .filter(|r| match r {
                Ok(_) => true,
                Err(e) => {
-                    error!(e; "export database job failed");
+                    error!(e; "export job failed");
                    false
                }
            })
-            .count();
-        let elapsed = timer.elapsed();
-
-        info!("Success {success}/{db_count} jobs, costs: {elapsed:?}");
-
-        Ok(())
+            .count()
    }
 }

@@ -493,3 +679,15 @@ impl Tool for Export {
        }
    }
 }
+
+/// Builds the WITH options string for SQL commands, assuming consistent syntax across S3 and local exports.
+fn build_with_options(start_time: &Option<String>, end_time: &Option<String>) -> String {
+    let mut options = vec!["format = 'parquet'".to_string()];
+    if let Some(start) = start_time {
+        options.push(format!("start_time = '{}'", start));
+    }
+    if let Some(end) = end_time {
+        options.push(format!("end_time = '{}'", end));
+    }
+    options.join(", ")
+}
--- a/src/cmd/src/datanode.rs
+++ b/src/cmd/src/datanode.rs
@@ -406,7 +406,7 @@ mod tests {
            sync_write = false

            [storage]
-            data_home = "/tmp/greptimedb/"
+            data_home = "./greptimedb_data/"
            type = "File"

            [[storage.providers]]
@@ -420,7 +420,7 @@ mod tests {

            [logging]
            level = "debug"
-            dir = "/tmp/greptimedb/test/logs"
+            dir = "./greptimedb_data/test/logs"
        "#;
        write!(file, "{}", toml_str).unwrap();

@@ -467,7 +467,7 @@ mod tests {
        assert_eq!(10000, ddl_timeout.as_millis());
        assert_eq!(3000, timeout.as_millis());
        assert!(tcp_nodelay);
-        assert_eq!("/tmp/greptimedb/", options.storage.data_home);
+        assert_eq!("./greptimedb_data/", options.storage.data_home);
        assert!(matches!(
            &options.storage.store,
            ObjectStoreConfig::File(FileConfig { .. })
@@ -483,7 +483,10 @@ mod tests {
        ));

        assert_eq!("debug", options.logging.level.unwrap());
-        assert_eq!("/tmp/greptimedb/test/logs".to_string(), options.logging.dir);
+        assert_eq!(
+            "./greptimedb_data/test/logs".to_string(),
+            options.logging.dir
+        );
    }

    #[test]
@@ -526,7 +529,7 @@ mod tests {

        let options = cmd
            .load_options(&GlobalOptions {
-                log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
+                log_dir: Some("./greptimedb_data/test/logs".to_string()),
                log_level: Some("debug".to_string()),

                #[cfg(feature = "tokio-console")]
@@ -536,7 +539,7 @@ mod tests {
            .component;

        let logging_opt = options.logging;
-        assert_eq!("/tmp/greptimedb/test/logs", logging_opt.dir);
+        assert_eq!("./greptimedb_data/test/logs", logging_opt.dir);
        assert_eq!("debug", logging_opt.level.as_ref().unwrap());
    }

@@ -565,11 +568,11 @@ mod tests {

            [storage]
            type = "File"
-            data_home = "/tmp/greptimedb/"
+            data_home = "./greptimedb_data/"

            [logging]
            level = "debug"
-            dir = "/tmp/greptimedb/test/logs"
+            dir = "./greptimedb_data/test/logs"
        "#;
        write!(file, "{}", toml_str).unwrap();

--- a/src/cmd/src/frontend.rs
+++ b/src/cmd/src/frontend.rs
@@ -448,7 +448,7 @@ mod tests {

            [logging]
            level = "debug"
-            dir = "/tmp/greptimedb/test/logs"
+            dir = "./greptimedb_data/test/logs"
        "#;
        write!(file, "{}", toml_str).unwrap();

@@ -466,7 +466,10 @@ mod tests {
        assert_eq!(ReadableSize::gb(2), fe_opts.http.body_limit);

        assert_eq!("debug", fe_opts.logging.level.as_ref().unwrap());
-        assert_eq!("/tmp/greptimedb/test/logs".to_string(), fe_opts.logging.dir);
+        assert_eq!(
+            "./greptimedb_data/test/logs".to_string(),
+            fe_opts.logging.dir
+        );
        assert!(!fe_opts.opentsdb.enable);
    }

@@ -505,7 +508,7 @@ mod tests {

        let options = cmd
            .load_options(&GlobalOptions {
-                log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
+                log_dir: Some("./greptimedb_data/test/logs".to_string()),
                log_level: Some("debug".to_string()),

                #[cfg(feature = "tokio-console")]
@@ -515,7 +518,7 @@ mod tests {
            .component;

        let logging_opt = options.logging;
-        assert_eq!("/tmp/greptimedb/test/logs", logging_opt.dir);
+        assert_eq!("./greptimedb_data/test/logs", logging_opt.dir);
        assert_eq!("debug", logging_opt.level.as_ref().unwrap());
    }

--- a/src/cmd/src/metasrv.rs
+++ b/src/cmd/src/metasrv.rs
@@ -337,7 +337,7 @@ mod tests {

            [logging]
            level = "debug"
-            dir = "/tmp/greptimedb/test/logs"
+            dir = "./greptimedb_data/test/logs"

            [failure_detector]
            threshold = 8.0
@@ -358,7 +358,10 @@ mod tests {
        assert_eq!(vec!["127.0.0.1:2379".to_string()], options.store_addrs);
        assert_eq!(SelectorType::LeaseBased, options.selector);
        assert_eq!("debug", options.logging.level.as_ref().unwrap());
-        assert_eq!("/tmp/greptimedb/test/logs".to_string(), options.logging.dir);
+        assert_eq!(
+            "./greptimedb_data/test/logs".to_string(),
+            options.logging.dir
+        );
        assert_eq!(8.0, options.failure_detector.threshold);
        assert_eq!(
            100.0,
@@ -396,7 +399,7 @@ mod tests {

        let options = cmd
            .load_options(&GlobalOptions {
-                log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
+                log_dir: Some("./greptimedb_data/test/logs".to_string()),
                log_level: Some("debug".to_string()),

                #[cfg(feature = "tokio-console")]
@@ -406,7 +409,7 @@ mod tests {
            .component;

        let logging_opt = options.logging;
-        assert_eq!("/tmp/greptimedb/test/logs", logging_opt.dir);
+        assert_eq!("./greptimedb_data/test/logs", logging_opt.dir);
        assert_eq!("debug", logging_opt.level.as_ref().unwrap());
    }

@@ -424,7 +427,7 @@ mod tests {

            [logging]
            level = "debug"
-            dir = "/tmp/greptimedb/test/logs"
+            dir = "./greptimedb_data/test/logs"
        "#;
        write!(file, "{}", toml_str).unwrap();

--- a/src/cmd/src/standalone.rs
+++ b/src/cmd/src/standalone.rs
@@ -852,7 +852,7 @@ mod tests {

            [wal]
            provider = "raft_engine"
-            dir = "/tmp/greptimedb/test/wal"
+            dir = "./greptimedb_data/test/wal"
            file_size = "1GB"
            purge_threshold = "50GB"
            purge_interval = "10m"
@@ -860,7 +860,7 @@ mod tests {
            sync_write = false

            [storage]
-            data_home = "/tmp/greptimedb/"
+            data_home = "./greptimedb_data/"
            type = "File"

            [[storage.providers]]
@@ -892,7 +892,7 @@ mod tests {

            [logging]
            level = "debug"
-            dir = "/tmp/greptimedb/test/logs"
+            dir = "./greptimedb_data/test/logs"
        "#;
        write!(file, "{}", toml_str).unwrap();
        let cmd = StartCommand {
@@ -922,7 +922,10 @@ mod tests {
        let DatanodeWalConfig::RaftEngine(raft_engine_config) = dn_opts.wal else {
            unreachable!()
        };
-        assert_eq!("/tmp/greptimedb/test/wal", raft_engine_config.dir.unwrap());
+        assert_eq!(
+            "./greptimedb_data/test/wal",
+            raft_engine_config.dir.unwrap()
+        );

        assert!(matches!(
            &dn_opts.storage.store,
@@ -946,7 +949,7 @@ mod tests {
        }

        assert_eq!("debug", logging_opts.level.as_ref().unwrap());
-        assert_eq!("/tmp/greptimedb/test/logs".to_string(), logging_opts.dir);
+        assert_eq!("./greptimedb_data/test/logs".to_string(), logging_opts.dir);
    }

    #[test]
@@ -958,7 +961,7 @@ mod tests {

        let opts = cmd
            .load_options(&GlobalOptions {
-                log_dir: Some("/tmp/greptimedb/test/logs".to_string()),
+                log_dir: Some("./greptimedb_data/test/logs".to_string()),
                log_level: Some("debug".to_string()),

                #[cfg(feature = "tokio-console")]
@@ -967,7 +970,7 @@ mod tests {
            .unwrap()
            .component;

-        assert_eq!("/tmp/greptimedb/test/logs", opts.logging.dir);
+        assert_eq!("./greptimedb_data/test/logs", opts.logging.dir);
        assert_eq!("debug", opts.logging.level.unwrap());
    }

--- a/src/cmd/tests/load_config_test.rs
+++ b/src/cmd/tests/load_config_test.rs
@@ -56,13 +56,13 @@ fn test_load_datanode_example_config() {
                metadata_cache_tti: Duration::from_secs(300),
            }),
            wal: DatanodeWalConfig::RaftEngine(RaftEngineConfig {
-                dir: Some("/tmp/greptimedb/wal".to_string()),
+                dir: Some("./greptimedb_data/wal".to_string()),
                sync_period: Some(Duration::from_secs(10)),
                recovery_parallelism: 2,
                ..Default::default()
            }),
            storage: StorageConfig {
-                data_home: "/tmp/greptimedb/".to_string(),
+                data_home: "./greptimedb_data/".to_string(),
                ..Default::default()
            },
            region_engine: vec![
@@ -159,10 +159,10 @@ fn test_load_metasrv_example_config() {
    let expected = GreptimeOptions::<MetasrvOptions> {
        component: MetasrvOptions {
            selector: SelectorType::default(),
-            data_home: "/tmp/metasrv/".to_string(),
+            data_home: "./greptimedb_data/metasrv/".to_string(),
            server_addr: "127.0.0.1:3002".to_string(),
            logging: LoggingOptions {
-                dir: "/tmp/greptimedb/logs".to_string(),
+                dir: "./greptimedb_data/logs".to_string(),
                level: Some("info".to_string()),
                otlp_endpoint: Some(DEFAULT_OTLP_ENDPOINT.to_string()),
                tracing_sample_ratio: Some(Default::default()),
@@ -202,7 +202,7 @@ fn test_load_standalone_example_config() {
        component: StandaloneOptions {
            default_timezone: Some("UTC".to_string()),
            wal: DatanodeWalConfig::RaftEngine(RaftEngineConfig {
-                dir: Some("/tmp/greptimedb/wal".to_string()),
+                dir: Some("./greptimedb_data/wal".to_string()),
                sync_period: Some(Duration::from_secs(10)),
                recovery_parallelism: 2,
                ..Default::default()
@@ -219,7 +219,7 @@ fn test_load_standalone_example_config() {
                }),
            ],
            storage: StorageConfig {
-                data_home: "/tmp/greptimedb/".to_string(),
+                data_home: "./greptimedb_data/".to_string(),
                ..Default::default()
            },
            logging: LoggingOptions {
--- a/src/common/config/src/config.rs
+++ b/src/common/config/src/config.rs
@@ -161,7 +161,7 @@ mod tests {

            [wal]
            provider = "raft_engine"
-            dir = "/tmp/greptimedb/wal"
+            dir = "./greptimedb_data/wal"
            file_size = "1GB"
            purge_threshold = "50GB"
            purge_interval = "10m"
@@ -170,7 +170,7 @@ mod tests {

            [logging]
            level = "debug"
-            dir = "/tmp/greptimedb/test/logs"
+            dir = "./greptimedb_data/test/logs"
        "#;
        write!(file, "{}", toml_str).unwrap();

@@ -246,7 +246,7 @@ mod tests {
                let DatanodeWalConfig::RaftEngine(raft_engine_config) = opts.wal else {
                    unreachable!()
                };
-                assert_eq!(raft_engine_config.dir.unwrap(), "/tmp/greptimedb/wal");
+                assert_eq!(raft_engine_config.dir.unwrap(), "./greptimedb_data/wal");

                // Should be default values.
                assert_eq!(opts.node_id, None);
--- a/src/common/function/Cargo.toml
+++ b/src/common/function/Cargo.toml
@@ -17,6 +17,7 @@ api.workspace = true
 arc-swap = "1.0"
 async-trait.workspace = true
 bincode = "1.3"
+chrono.workspace = true
 common-base.workspace = true
 common-catalog.workspace = true
 common-error.workspace = true
--- a/src/common/function/src/scalars/date/date_format.rs
+++ b/src/common/function/src/scalars/date/date_format.rs
@@ -43,7 +43,6 @@ impl Function for DateFormatFunction {
        helper::one_of_sigs2(
            vec![
                ConcreteDataType::date_datatype(),
-                ConcreteDataType::datetime_datatype(),
                ConcreteDataType::timestamp_second_datatype(),
                ConcreteDataType::timestamp_millisecond_datatype(),
                ConcreteDataType::timestamp_microsecond_datatype(),
@@ -105,22 +104,6 @@ impl Function for DateFormatFunction {
                    results.push(result.as_deref());
                }
            }
-            ConcreteDataType::DateTime(_) => {
-                for i in 0..size {
-                    let datetime = left.get(i).as_datetime();
-                    let format = formats.get(i).as_string();
-
-                    let result = match (datetime, format) {
-                        (Some(datetime), Some(fmt)) => datetime
-                            .as_formatted_string(&fmt, Some(&func_ctx.query_ctx.timezone()))
-                            .map_err(BoxedError::new)
-                            .context(error::ExecuteSnafu)?,
-                        _ => None,
-                    };
-
-                    results.push(result.as_deref());
-                }
-            }
            _ => {
                return UnsupportedInputDataTypeSnafu {
                    function: NAME,
@@ -147,7 +130,7 @@ mod tests {
    use common_query::prelude::{TypeSignature, Volatility};
    use datatypes::prelude::{ConcreteDataType, ScalarVector};
    use datatypes::value::Value;
-    use datatypes::vectors::{DateTimeVector, DateVector, StringVector, TimestampSecondVector};
+    use datatypes::vectors::{DateVector, StringVector, TimestampSecondVector};

    use super::{DateFormatFunction, *};

@@ -169,16 +152,11 @@ mod tests {
            ConcreteDataType::string_datatype(),
            f.return_type(&[ConcreteDataType::date_datatype()]).unwrap()
        );
-        assert_eq!(
-            ConcreteDataType::string_datatype(),
-            f.return_type(&[ConcreteDataType::datetime_datatype()])
-                .unwrap()
-        );
        assert!(matches!(f.signature(),
                         Signature {
                             type_signature: TypeSignature::OneOf(sigs),
                             volatility: Volatility::Immutable
-                         } if  sigs.len() == 6));
+                         } if  sigs.len() == 5));
    }

    #[test]
@@ -262,45 +240,4 @@ mod tests {
            }
        }
    }
-
-    #[test]
-    fn test_datetime_date_format() {
-        let f = DateFormatFunction;
-
-        let dates = vec![Some(123), None, Some(42), None];
-        let formats = vec![
-            "%Y-%m-%d %T.%3f",
-            "%Y-%m-%d %T.%3f",
-            "%Y-%m-%d %T.%3f",
-            "%Y-%m-%d %T.%3f",
-        ];
-        let results = [
-            Some("1970-01-01 00:00:00.123"),
-            None,
-            Some("1970-01-01 00:00:00.042"),
-            None,
-        ];
-
-        let date_vector = DateTimeVector::from(dates.clone());
-        let interval_vector = StringVector::from_vec(formats);
-        let args: Vec<VectorRef> = vec![Arc::new(date_vector), Arc::new(interval_vector)];
-        let vector = f.eval(&FunctionContext::default(), &args).unwrap();
-
-        assert_eq!(4, vector.len());
-        for (i, _t) in dates.iter().enumerate() {
-            let v = vector.get(i);
-            let result = results.get(i).unwrap();
-
-            if result.is_none() {
-                assert_eq!(Value::Null, v);
-                continue;
-            }
-            match v {
-                Value::String(s) => {
-                    assert_eq!(s.as_utf8(), result.unwrap());
-                }
-                _ => unreachable!(),
-            }
-        }
-    }
 }
--- a/src/common/function/src/scalars/date/date_sub.rs
+++ b/src/common/function/src/scalars/date/date_sub.rs
@@ -118,11 +118,6 @@ mod tests {
            ConcreteDataType::date_datatype(),
            f.return_type(&[ConcreteDataType::date_datatype()]).unwrap()
        );
-        assert_eq!(
-            ConcreteDataType::datetime_datatype(),
-            f.return_type(&[ConcreteDataType::datetime_datatype()])
-                .unwrap()
-        );
        assert!(
            matches!(f.signature(),
                         Signature {
--- a/src/common/function/src/scalars/timestamp/greatest.rs
+++ b/src/common/function/src/scalars/timestamp/greatest.rs
@@ -23,7 +23,7 @@ use datatypes::arrow::array::AsArray;
 use datatypes::arrow::compute::cast;
 use datatypes::arrow::compute::kernels::zip;
 use datatypes::arrow::datatypes::{
-    DataType as ArrowDataType, Date32Type, Date64Type, TimestampMicrosecondType,
+    DataType as ArrowDataType, Date32Type, TimeUnit, TimestampMicrosecondType,
    TimestampMillisecondType, TimestampNanosecondType, TimestampSecondType,
 };
 use datatypes::prelude::ConcreteDataType;
@@ -69,9 +69,8 @@ impl Function for GreatestFunction {
        );

        match &input_types[0] {
-            ConcreteDataType::String(_) => Ok(ConcreteDataType::datetime_datatype()),
+            ConcreteDataType::String(_) => Ok(ConcreteDataType::timestamp_millisecond_datatype()),
            ConcreteDataType::Date(_) => Ok(ConcreteDataType::date_datatype()),
-            ConcreteDataType::DateTime(_) => Ok(ConcreteDataType::datetime_datatype()),
            ConcreteDataType::Timestamp(ts_type) => Ok(ConcreteDataType::Timestamp(*ts_type)),
            _ => UnsupportedInputDataTypeSnafu {
                function: NAME,
@@ -87,7 +86,6 @@ impl Function for GreatestFunction {
            vec![
                ConcreteDataType::string_datatype(),
                ConcreteDataType::date_datatype(),
-                ConcreteDataType::datetime_datatype(),
                ConcreteDataType::timestamp_nanosecond_datatype(),
                ConcreteDataType::timestamp_microsecond_datatype(),
                ConcreteDataType::timestamp_millisecond_datatype(),
@@ -109,20 +107,24 @@ impl Function for GreatestFunction {
        );
        match columns[0].data_type() {
            ConcreteDataType::String(_) => {
-                // Treats string as `DateTime` type.
-                let column1 = cast(&columns[0].to_arrow_array(), &ArrowDataType::Date64)
-                    .context(ArrowComputeSnafu)?;
-                let column1 = column1.as_primitive::<Date64Type>();
-                let column2 = cast(&columns[1].to_arrow_array(), &ArrowDataType::Date64)
-                    .context(ArrowComputeSnafu)?;
-                let column2 = column2.as_primitive::<Date64Type>();
+                let column1 = cast(
+                    &columns[0].to_arrow_array(),
+                    &ArrowDataType::Timestamp(TimeUnit::Millisecond, None),
+                )
+                .context(ArrowComputeSnafu)?;
+                let column1 = column1.as_primitive::<TimestampMillisecondType>();
+                let column2 = cast(
+                    &columns[1].to_arrow_array(),
+                    &ArrowDataType::Timestamp(TimeUnit::Millisecond, None),
+                )
+                .context(ArrowComputeSnafu)?;
+                let column2 = column2.as_primitive::<TimestampMillisecondType>();
                let boolean_array = gt(&column1, &column2).context(ArrowComputeSnafu)?;
                let result =
                    zip::zip(&boolean_array, &column1, &column2).context(ArrowComputeSnafu)?;
                Ok(Helper::try_into_vector(&result).context(error::FromArrowArraySnafu)?)
            }
            ConcreteDataType::Date(_) => gt_time_types!(Date32Type, columns),
-            ConcreteDataType::DateTime(_) => gt_time_types!(Date64Type, columns),
            ConcreteDataType::Timestamp(ts_type) => match ts_type {
                TimestampType::Second(_) => gt_time_types!(TimestampSecondType, columns),
                TimestampType::Millisecond(_) => {
@@ -155,15 +157,15 @@ mod tests {
    use std::sync::Arc;

    use common_time::timestamp::TimeUnit;
-    use common_time::{Date, DateTime, Timestamp};
+    use common_time::{Date, Timestamp};
    use datatypes::types::{
-        DateTimeType, DateType, TimestampMicrosecondType, TimestampMillisecondType,
-        TimestampNanosecondType, TimestampSecondType,
+        DateType, TimestampMicrosecondType, TimestampMillisecondType, TimestampNanosecondType,
+        TimestampSecondType,
    };
    use datatypes::value::Value;
    use datatypes::vectors::{
-        DateTimeVector, DateVector, StringVector, TimestampMicrosecondVector,
-        TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, Vector,
+        DateVector, StringVector, TimestampMicrosecondVector, TimestampMillisecondVector,
+        TimestampNanosecondVector, TimestampSecondVector, Vector,
    };
    use paste::paste;

@@ -178,7 +180,7 @@ mod tests {
                    ConcreteDataType::string_datatype()
                ])
                .unwrap(),
-            ConcreteDataType::DateTime(DateTimeType)
+            ConcreteDataType::timestamp_millisecond_datatype()
        );
        let columns = vec![
            Arc::new(StringVector::from(vec![
@@ -194,15 +196,18 @@ mod tests {
        let result = function
            .eval(&FunctionContext::default(), &columns)
            .unwrap();
-        let result = result.as_any().downcast_ref::<DateTimeVector>().unwrap();
+        let result = result
+            .as_any()
+            .downcast_ref::<TimestampMillisecondVector>()
+            .unwrap();
        assert_eq!(result.len(), 2);
        assert_eq!(
            result.get(0),
-            Value::DateTime(DateTime::from_str("2001-02-01 00:00:00", None).unwrap())
+            Value::Timestamp(Timestamp::from_str("2001-02-01 00:00:00", None).unwrap())
        );
        assert_eq!(
            result.get(1),
-            Value::DateTime(DateTime::from_str("2012-12-23 00:00:00", None).unwrap())
+            Value::Timestamp(Timestamp::from_str("2012-12-23 00:00:00", None).unwrap())
        );
    }

@@ -245,30 +250,33 @@ mod tests {
        assert_eq!(
            function
                .return_type(&[
-                    ConcreteDataType::datetime_datatype(),
-                    ConcreteDataType::datetime_datatype()
+                    ConcreteDataType::timestamp_millisecond_datatype(),
+                    ConcreteDataType::timestamp_millisecond_datatype()
                ])
                .unwrap(),
-            ConcreteDataType::DateTime(DateTimeType)
+            ConcreteDataType::timestamp_millisecond_datatype()
        );

        let columns = vec![
-            Arc::new(DateTimeVector::from_slice(vec![-1, 2])) as _,
-            Arc::new(DateTimeVector::from_slice(vec![0, 1])) as _,
+            Arc::new(TimestampMillisecondVector::from_slice(vec![-1, 2])) as _,
+            Arc::new(TimestampMillisecondVector::from_slice(vec![0, 1])) as _,
        ];

        let result = function
            .eval(&FunctionContext::default(), &columns)
            .unwrap();
-        let result = result.as_any().downcast_ref::<DateTimeVector>().unwrap();
+        let result = result
+            .as_any()
+            .downcast_ref::<TimestampMillisecondVector>()
+            .unwrap();
        assert_eq!(result.len(), 2);
        assert_eq!(
            result.get(0),
-            Value::DateTime(DateTime::from_str("1970-01-01 00:00:00", None).unwrap())
+            Value::Timestamp(Timestamp::from_str("1970-01-01 00:00:00", None).unwrap())
        );
        assert_eq!(
            result.get(1),
-            Value::DateTime(DateTime::from_str("1970-01-01 00:00:00.002", None).unwrap())
+            Value::Timestamp(Timestamp::from_str("1970-01-01 00:00:00.002", None).unwrap())
        );
    }

--- a/src/common/function/src/scalars/timestamp/to_unixtime.rs
+++ b/src/common/function/src/scalars/timestamp/to_unixtime.rs
@@ -17,7 +17,7 @@ use std::sync::Arc;

 use common_query::error::{InvalidFuncArgsSnafu, Result, UnsupportedInputDataTypeSnafu};
 use common_query::prelude::{Signature, Volatility};
-use common_time::{Date, DateTime, Timestamp};
+use common_time::{Date, Timestamp};
 use datatypes::prelude::ConcreteDataType;
 use datatypes::vectors::{Int64Vector, VectorRef};
 use snafu::ensure;
@@ -32,10 +32,6 @@ const NAME: &str = "to_unixtime";

 fn convert_to_seconds(arg: &str, func_ctx: &FunctionContext) -> Option<i64> {
    let timezone = &func_ctx.query_ctx.timezone();
-    if let Ok(dt) = DateTime::from_str(arg, Some(timezone)) {
-        return Some(dt.val() / 1000);
-    }
-
    if let Ok(ts) = Timestamp::from_str(arg, Some(timezone)) {
        return Some(ts.split().0);
    }
@@ -59,12 +55,6 @@ fn convert_dates_to_seconds(vector: &VectorRef) -> Vec<Option<i64>> {
        .collect::<Vec<Option<i64>>>()
 }

-fn convert_datetimes_to_seconds(vector: &VectorRef) -> Vec<Option<i64>> {
-    (0..vector.len())
-        .map(|i| vector.get(i).as_datetime().map(|dt| dt.val() / 1000))
-        .collect::<Vec<Option<i64>>>()
-}
-
 impl Function for ToUnixtimeFunction {
    fn name(&self) -> &str {
        NAME
@@ -82,7 +72,6 @@ impl Function for ToUnixtimeFunction {
                ConcreteDataType::int32_datatype(),
                ConcreteDataType::int64_datatype(),
                ConcreteDataType::date_datatype(),
-                ConcreteDataType::datetime_datatype(),
                ConcreteDataType::timestamp_second_datatype(),
                ConcreteDataType::timestamp_millisecond_datatype(),
                ConcreteDataType::timestamp_microsecond_datatype(),
@@ -119,10 +108,6 @@ impl Function for ToUnixtimeFunction {
                let seconds = convert_dates_to_seconds(vector);
                Ok(Arc::new(Int64Vector::from(seconds)))
            }
-            ConcreteDataType::DateTime(_) => {
-                let seconds = convert_datetimes_to_seconds(vector);
-                Ok(Arc::new(Int64Vector::from(seconds)))
-            }
            ConcreteDataType::Timestamp(_) => {
                let seconds = convert_timestamps_to_seconds(vector);
                Ok(Arc::new(Int64Vector::from(seconds)))
@@ -148,7 +133,7 @@ mod tests {
    use datatypes::prelude::ConcreteDataType;
    use datatypes::value::Value;
    use datatypes::vectors::{
-        DateTimeVector, DateVector, StringVector, TimestampMillisecondVector, TimestampSecondVector,
+        DateVector, StringVector, TimestampMillisecondVector, TimestampSecondVector,
    };

    use super::{ToUnixtimeFunction, *};
@@ -171,7 +156,6 @@ mod tests {
                             ConcreteDataType::int32_datatype(),
                             ConcreteDataType::int64_datatype(),
                             ConcreteDataType::date_datatype(),
-                             ConcreteDataType::datetime_datatype(),
                             ConcreteDataType::timestamp_second_datatype(),
                             ConcreteDataType::timestamp_millisecond_datatype(),
                             ConcreteDataType::timestamp_microsecond_datatype(),
@@ -253,31 +237,6 @@ mod tests {
        }
    }

-    #[test]
-    fn test_datetime_to_unixtime() {
-        let f = ToUnixtimeFunction;
-
-        let times = vec![Some(123000), None, Some(42000), None];
-        let results = [Some(123), None, Some(42), None];
-        let date_vector = DateTimeVector::from(times.clone());
-        let args: Vec<VectorRef> = vec![Arc::new(date_vector)];
-        let vector = f.eval(&FunctionContext::default(), &args).unwrap();
-        assert_eq!(4, vector.len());
-        for (i, _t) in times.iter().enumerate() {
-            let v = vector.get(i);
-            if i == 1 || i == 3 {
-                assert_eq!(Value::Null, v);
-                continue;
-            }
-            match v {
-                Value::Int64(ts) => {
-                    assert_eq!(ts, (*results.get(i).unwrap()).unwrap());
-                }
-                _ => unreachable!(),
-            }
-        }
-    }
-
    #[test]
    fn test_timestamp_to_unixtime() {
        let f = ToUnixtimeFunction;
--- a/src/common/function/src/scalars/vector.rs
+++ b/src/common/function/src/scalars/vector.rs
@@ -27,6 +27,7 @@ mod vector_div;
 mod vector_mul;
 mod vector_norm;
 mod vector_sub;
+mod vector_subvector;

 use std::sync::Arc;

@@ -56,6 +57,7 @@ impl VectorFunction {
        registry.register(Arc::new(vector_div::VectorDivFunction));
        registry.register(Arc::new(vector_norm::VectorNormFunction));
        registry.register(Arc::new(vector_dim::VectorDimFunction));
+        registry.register(Arc::new(vector_subvector::VectorSubvectorFunction));
        registry.register(Arc::new(elem_sum::ElemSumFunction));
        registry.register(Arc::new(elem_product::ElemProductFunction));
    }
--- a/src/common/function/src/scalars/vector/vector_subvector.rs
+++ b/src/common/function/src/scalars/vector/vector_subvector.rs
@@ -0,0 +1,240 @@
+// 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 std::fmt::Display;
+
+use common_query::error::{InvalidFuncArgsSnafu, Result};
+use common_query::prelude::{Signature, TypeSignature};
+use datafusion_expr::Volatility;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BinaryVectorBuilder, MutableVector, VectorRef};
+use snafu::ensure;
+
+use crate::function::{Function, FunctionContext};
+use crate::scalars::vector::impl_conv::{as_veclit, as_veclit_if_const, veclit_to_binlit};
+
+const NAME: &str = "vec_subvector";
+
+/// Returns a subvector from start(included) to end(excluded) index.
+///
+/// # Example
+///
+/// ```sql
+/// SELECT vec_to_string(vec_subvector("[1, 2, 3, 4, 5]", 1, 3)) as result;
+///
+/// +---------+
+/// | result  |
+/// +---------+
+/// | [2, 3]  |
+/// +---------+
+///
+/// ```
+///
+
+#[derive(Debug, Clone, Default)]
+pub struct VectorSubvectorFunction;
+
+impl Function for VectorSubvectorFunction {
+    fn name(&self) -> &str {
+        NAME
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::binary_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::one_of(
+            vec![
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::string_datatype(),
+                    ConcreteDataType::int64_datatype(),
+                    ConcreteDataType::int64_datatype(),
+                ]),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::binary_datatype(),
+                    ConcreteDataType::int64_datatype(),
+                    ConcreteDataType::int64_datatype(),
+                ]),
+            ],
+            Volatility::Immutable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: &FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure!(
+            columns.len() == 3,
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The length of the args is not correct, expect exactly three, have: {}",
+                    columns.len()
+                )
+            }
+        );
+
+        let arg0 = &columns[0];
+        let arg1 = &columns[1];
+        let arg2 = &columns[2];
+
+        ensure!(
+            arg0.len() == arg1.len() && arg1.len() == arg2.len(),
+            InvalidFuncArgsSnafu {
+                err_msg: format!(
+                    "The lengths of the vector are not aligned, args 0: {}, args 1: {}, args 2: {}",
+                    arg0.len(),
+                    arg1.len(),
+                    arg2.len()
+                )
+            }
+        );
+
+        let len = arg0.len();
+        let mut result = BinaryVectorBuilder::with_capacity(len);
+        if len == 0 {
+            return Ok(result.to_vector());
+        }
+
+        let arg0_const = as_veclit_if_const(arg0)?;
+
+        for i in 0..len {
+            let arg0 = match arg0_const.as_ref() {
+                Some(arg0) => Some(Cow::Borrowed(arg0.as_ref())),
+                None => as_veclit(arg0.get_ref(i))?,
+            };
+            let arg1 = arg1.get(i).as_i64();
+            let arg2 = arg2.get(i).as_i64();
+            let (Some(arg0), Some(arg1), Some(arg2)) = (arg0, arg1, arg2) else {
+                result.push_null();
+                continue;
+            };
+
+            ensure!(
+                0 <= arg1 && arg1 <= arg2 && arg2 as usize <= arg0.len(),
+                InvalidFuncArgsSnafu {
+                    err_msg: format!(
+                        "Invalid start and end indices: start={}, end={}, vec_len={}",
+                        arg1,
+                        arg2,
+                        arg0.len()
+                    )
+                }
+            );
+
+            let subvector = &arg0[arg1 as usize..arg2 as usize];
+            let binlit = veclit_to_binlit(subvector);
+            result.push(Some(&binlit));
+        }
+
+        Ok(result.to_vector())
+    }
+}
+
+impl Display for VectorSubvectorFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", NAME.to_ascii_uppercase())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use common_query::error::Error;
+    use datatypes::vectors::{Int64Vector, StringVector};
+
+    use super::*;
+    use crate::function::FunctionContext;
+    #[test]
+    fn test_subvector() {
+        let func = VectorSubvectorFunction;
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0, 2.0, 3.0, 4.0, 5.0]".to_string()),
+            Some("[6.0, 7.0, 8.0, 9.0, 10.0]".to_string()),
+            None,
+            Some("[11.0, 12.0, 13.0]".to_string()),
+        ]));
+        let input1 = Arc::new(Int64Vector::from(vec![Some(1), Some(0), Some(0), Some(1)]));
+        let input2 = Arc::new(Int64Vector::from(vec![Some(3), Some(5), Some(2), Some(3)]));
+
+        let result = func
+            .eval(&FunctionContext::default(), &[input0, input1, input2])
+            .unwrap();
+
+        let result = result.as_ref();
+        assert_eq!(result.len(), 4);
+        assert_eq!(
+            result.get_ref(0).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[2.0, 3.0]).as_slice())
+        );
+        assert_eq!(
+            result.get_ref(1).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[6.0, 7.0, 8.0, 9.0, 10.0]).as_slice())
+        );
+        assert!(result.get_ref(2).is_null());
+        assert_eq!(
+            result.get_ref(3).as_binary().unwrap(),
+            Some(veclit_to_binlit(&[12.0, 13.0]).as_slice())
+        );
+    }
+    #[test]
+    fn test_subvector_error() {
+        let func = VectorSubvectorFunction;
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0, 2.0, 3.0]".to_string()),
+            Some("[4.0, 5.0, 6.0]".to_string()),
+        ]));
+        let input1 = Arc::new(Int64Vector::from(vec![Some(1), Some(2)]));
+        let input2 = Arc::new(Int64Vector::from(vec![Some(3)]));
+
+        let result = func.eval(&FunctionContext::default(), &[input0, input1, input2]);
+
+        match result {
+            Err(Error::InvalidFuncArgs { err_msg, .. }) => {
+                assert_eq!(
+                    err_msg,
+                    "The lengths of the vector are not aligned, args 0: 2, args 1: 2, args 2: 1"
+                )
+            }
+            _ => unreachable!(),
+        }
+    }
+
+    #[test]
+    fn test_subvector_invalid_indices() {
+        let func = VectorSubvectorFunction;
+
+        let input0 = Arc::new(StringVector::from(vec![
+            Some("[1.0, 2.0, 3.0]".to_string()),
+            Some("[4.0, 5.0, 6.0]".to_string()),
+        ]));
+        let input1 = Arc::new(Int64Vector::from(vec![Some(1), Some(3)]));
+        let input2 = Arc::new(Int64Vector::from(vec![Some(3), Some(4)]));
+
+        let result = func.eval(&FunctionContext::default(), &[input0, input1, input2]);
+
+        match result {
+            Err(Error::InvalidFuncArgs { err_msg, .. }) => {
+                assert_eq!(
+                    err_msg,
+                    "Invalid start and end indices: start=3, end=4, vec_len=3"
+                )
+            }
+            _ => unreachable!(),
+        }
+    }
+}
--- a/src/common/grpc/src/select.rs
+++ b/src/common/grpc/src/select.rs
@@ -17,8 +17,8 @@ use api::v1::column::Values;
 use common_base::BitVec;
 use datatypes::types::{IntervalType, TimeType, TimestampType, WrapperType};
 use datatypes::vectors::{
-    BinaryVector, BooleanVector, DateTimeVector, DateVector, Decimal128Vector, Float32Vector,
-    Float64Vector, Int16Vector, Int32Vector, Int64Vector, Int8Vector, IntervalDayTimeVector,
+    BinaryVector, BooleanVector, DateVector, Decimal128Vector, Float32Vector, Float64Vector,
+    Int16Vector, Int32Vector, Int64Vector, Int8Vector, IntervalDayTimeVector,
    IntervalMonthDayNanoVector, IntervalYearMonthVector, StringVector, TimeMicrosecondVector,
    TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector,
    TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, UInt16Vector,
@@ -141,12 +141,6 @@ pub fn values(arrays: &[VectorRef]) -> Result<Values> {
        (ConcreteDataType::Date(_), DateVector, date_values, |x| {
            x.val()
        }),
-        (
-            ConcreteDataType::DateTime(_),
-            DateTimeVector,
-            datetime_values,
-            |x| { x.val() }
-        ),
        (
            ConcreteDataType::Timestamp(TimestampType::Second(_)),
            TimestampSecondVector,
--- a/src/common/macro/src/lib.rs
+++ b/src/common/macro/src/lib.rs
@@ -18,11 +18,13 @@ mod print_caller;
 mod range_fn;
 mod stack_trace_debug;
 mod utils;
+
 use aggr_func::{impl_aggr_func_type_store, impl_as_aggr_func_creator};
 use print_caller::process_print_caller;
 use proc_macro::TokenStream;
+use quote::quote;
 use range_fn::process_range_fn;
-use syn::{parse_macro_input, DeriveInput};
+use syn::{parse_macro_input, Data, DeriveInput, Fields};

 use crate::admin_fn::process_admin_fn;

@@ -136,3 +138,51 @@ pub fn print_caller(args: TokenStream, input: TokenStream) -> TokenStream {
 pub fn stack_trace_debug(args: TokenStream, input: TokenStream) -> TokenStream {
    stack_trace_debug::stack_trace_style_impl(args.into(), input.into()).into()
 }
+
+/// Generates implementation for `From<&TableMeta> for TableMetaBuilder`
+#[proc_macro_derive(ToMetaBuilder)]
+pub fn derive_meta_builder(input: TokenStream) -> TokenStream {
+    let input = parse_macro_input!(input as DeriveInput);
+
+    let Data::Struct(data_struct) = input.data else {
+        panic!("ToMetaBuilder can only be derived for structs");
+    };
+
+    let Fields::Named(fields) = data_struct.fields else {
+        panic!("ToMetaBuilder can only be derived for structs with named fields");
+    };
+
+    // Check that this is being applied to TableMeta struct
+    if input.ident != "TableMeta" {
+        panic!("ToMetaBuilder can only be derived for TableMeta struct");
+    }
+
+    let field_init = fields.named.iter().map(|field| {
+        let field_name = field.ident.as_ref().unwrap();
+        quote! {
+            #field_name: Default::default(),
+        }
+    });
+
+    let field_assignments = fields.named.iter().map(|field| {
+        let field_name = field.ident.as_ref().unwrap();
+        quote! {
+            builder.#field_name(meta.#field_name.clone());
+        }
+    });
+
+    let gen = quote! {
+        impl From<&TableMeta> for TableMetaBuilder {
+            fn from(meta: &TableMeta) -> Self {
+                let mut builder = Self {
+                    #(#field_init)*
+                };
+
+                #(#field_assignments)*
+                builder
+            }
+        }
+    };
+
+    gen.into()
+}
--- a/src/common/meta/Cargo.toml
+++ b/src/common/meta/Cargo.toml
@@ -7,6 +7,7 @@ license.workspace = true
 [features]
 testing = []
 pg_kvbackend = ["dep:tokio-postgres", "dep:backon", "dep:deadpool-postgres", "dep:deadpool"]
+mysql_kvbackend = ["dep:sqlx", "dep:backon"]

 [lints]
 workspace = true
@@ -57,9 +58,10 @@ serde_json.workspace = true
 serde_with.workspace = true
 session.workspace = true
 snafu.workspace = true
+sqlx = { workspace = true, optional = true }
 store-api.workspace = true
 strum.workspace = true
-table.workspace = true
+table = { workspace = true, features = ["testing"] }
 tokio.workspace = true
 tokio-postgres = { workspace = true, optional = true }
 tonic.workspace = true
--- a/src/common/meta/src/cache/flow/table_flownode.rs
+++ b/src/common/meta/src/cache/flow/table_flownode.rs
@@ -192,6 +192,8 @@ mod tests {
                    expire_after: Some(300),
                    comment: "comment".to_string(),
                    options: Default::default(),
+                    created_time: chrono::Utc::now(),
+                    updated_time: chrono::Utc::now(),
                },
                (1..=3)
                    .map(|i| {
--- a/src/common/meta/src/ddl/create_flow.rs
+++ b/src/common/meta/src/ddl/create_flow.rs
@@ -425,7 +425,14 @@ impl From<&CreateFlowData> for (FlowInfoValue, Vec<(FlowPartitionId, FlowRouteVa
        let flow_type = value.flow_type.unwrap_or_default().to_string();
        options.insert("flow_type".to_string(), flow_type);

-        let flow_info = FlowInfoValue {
+        let mut create_time = chrono::Utc::now();
+        if let Some(prev_flow_value) = value.prev_flow_info_value.as_ref()
+            && value.task.or_replace
+        {
+            create_time = prev_flow_value.get_inner_ref().created_time;
+        }
+
+        let flow_info: FlowInfoValue = FlowInfoValue {
            source_table_ids: value.source_table_ids.clone(),
            sink_table_name,
            flownode_ids,
@@ -435,6 +442,8 @@ impl From<&CreateFlowData> for (FlowInfoValue, Vec<(FlowPartitionId, FlowRouteVa
            expire_after,
            comment,
            options,
+            created_time: create_time,
+            updated_time: chrono::Utc::now(),
        };

        (flow_info, flow_routes)
--- a/src/common/meta/src/error.rs
+++ b/src/common/meta/src/error.rs
@@ -685,7 +685,36 @@ pub enum Error {
        operation: String,
    },

-    #[cfg(feature = "pg_kvbackend")]
+    #[cfg(feature = "mysql_kvbackend")]
+    #[snafu(display("Failed to execute via MySql, sql: {}", sql))]
+    MySqlExecution {
+        sql: String,
+        #[snafu(source)]
+        error: sqlx::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[cfg(feature = "mysql_kvbackend")]
+    #[snafu(display("Failed to create connection pool for MySql"))]
+    CreateMySqlPool {
+        #[snafu(source)]
+        error: sqlx::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[cfg(feature = "mysql_kvbackend")]
+    #[snafu(display("Failed to {} MySql transaction", operation))]
+    MySqlTransaction {
+        #[snafu(source)]
+        error: sqlx::Error,
+        #[snafu(implicit)]
+        location: Location,
+        operation: String,
+    },
+
+    #[cfg(any(feature = "pg_kvbackend", feature = "mysql_kvbackend"))]
    #[snafu(display("Rds transaction retry failed"))]
    RdsTransactionRetryFailed {
        #[snafu(implicit)]
@@ -823,8 +852,13 @@ impl ErrorExt for Error {
            PostgresExecution { .. }
            | CreatePostgresPool { .. }
            | GetPostgresConnection { .. }
-            | PostgresTransaction { .. }
-            | RdsTransactionRetryFailed { .. } => StatusCode::Internal,
+            | PostgresTransaction { .. } => StatusCode::Internal,
+            #[cfg(feature = "mysql_kvbackend")]
+            MySqlExecution { .. } | CreateMySqlPool { .. } | MySqlTransaction { .. } => {
+                StatusCode::Internal
+            }
+            #[cfg(any(feature = "pg_kvbackend", feature = "mysql_kvbackend"))]
+            RdsTransactionRetryFailed { .. } => StatusCode::Internal,
            Error::DatanodeTableInfoNotFound { .. } => StatusCode::Internal,
        }
    }
@@ -835,16 +869,29 @@ impl ErrorExt for Error {
 }

 impl Error {
-    #[cfg(feature = "pg_kvbackend")]
+    #[cfg(any(feature = "pg_kvbackend", feature = "mysql_kvbackend"))]
    /// Check if the error is a serialization error.
    pub fn is_serialization_error(&self) -> bool {
        match self {
+            #[cfg(feature = "pg_kvbackend")]
            Error::PostgresTransaction { error, .. } => {
                error.code() == Some(&tokio_postgres::error::SqlState::T_R_SERIALIZATION_FAILURE)
            }
+            #[cfg(feature = "pg_kvbackend")]
            Error::PostgresExecution { error, .. } => {
                error.code() == Some(&tokio_postgres::error::SqlState::T_R_SERIALIZATION_FAILURE)
            }
+            #[cfg(feature = "mysql_kvbackend")]
+            Error::MySqlExecution {
+                error: sqlx::Error::Database(database_error),
+                ..
+            } => {
+                matches!(
+                    database_error.message(),
+                    "Deadlock found when trying to get lock; try restarting transaction"
+                        | "can't serialize access for this transaction"
+                )
+            }
            _ => false,
        }
    }
--- a/src/common/meta/src/key/flow.rs
+++ b/src/common/meta/src/key/flow.rs
@@ -461,6 +461,8 @@ mod tests {
            expire_after: Some(300),
            comment: "hi".to_string(),
            options: Default::default(),
+            created_time: chrono::Utc::now(),
+            updated_time: chrono::Utc::now(),
        }
    }

@@ -632,6 +634,8 @@ mod tests {
            expire_after: Some(300),
            comment: "hi".to_string(),
            options: Default::default(),
+            created_time: chrono::Utc::now(),
+            updated_time: chrono::Utc::now(),
        };
        let err = flow_metadata_manager
            .create_flow_metadata(flow_id, flow_value, flow_routes.clone())
@@ -869,6 +873,8 @@ mod tests {
            expire_after: Some(300),
            comment: "hi".to_string(),
            options: Default::default(),
+            created_time: chrono::Utc::now(),
+            updated_time: chrono::Utc::now(),
        };
        let err = flow_metadata_manager
            .update_flow_metadata(
--- a/src/common/meta/src/key/flow/flow_info.rs
+++ b/src/common/meta/src/key/flow/flow_info.rs
@@ -15,6 +15,7 @@
 use std::collections::{BTreeMap, HashMap};
 use std::sync::Arc;

+use chrono::{DateTime, Utc};
 use lazy_static::lazy_static;
 use regex::Regex;
 use serde::{Deserialize, Serialize};
@@ -131,6 +132,12 @@ pub struct FlowInfoValue {
    pub(crate) comment: String,
    /// The options.
    pub(crate) options: HashMap<String, String>,
+    /// The created time
+    #[serde(default)]
+    pub(crate) created_time: DateTime<Utc>,
+    /// The updated time.
+    #[serde(default)]
+    pub(crate) updated_time: DateTime<Utc>,
 }

 impl FlowInfoValue {
@@ -171,6 +178,14 @@ impl FlowInfoValue {
    pub fn options(&self) -> &HashMap<String, String> {
        &self.options
    }
+
+    pub fn created_time(&self) -> &DateTime<Utc> {
+        &self.created_time
+    }
+
+    pub fn updated_time(&self) -> &DateTime<Utc> {
+        &self.updated_time
+    }
 }

 pub type FlowInfoManagerRef = Arc<FlowInfoManager>;
--- a/src/common/meta/src/key/flow/flow_state.rs
+++ b/src/common/meta/src/key/flow/flow_state.rs
@@ -97,11 +97,19 @@ impl<'a> MetadataKey<'a, FlowStateKey> for FlowStateKey {
 pub struct FlowStateValue {
    /// For each key, the bytes of the state in memory
    pub state_size: BTreeMap<FlowId, usize>,
+    /// For each key, the last execution time of flow in unix timestamp milliseconds.
+    pub last_exec_time_map: BTreeMap<FlowId, i64>,
 }

 impl FlowStateValue {
-    pub fn new(state_size: BTreeMap<FlowId, usize>) -> Self {
-        Self { state_size }
+    pub fn new(
+        state_size: BTreeMap<FlowId, usize>,
+        last_exec_time_map: BTreeMap<FlowId, i64>,
+    ) -> Self {
+        Self {
+            state_size,
+            last_exec_time_map,
+        }
    }
 }

@@ -143,12 +151,15 @@ impl FlowStateManager {
 pub struct FlowStat {
    /// For each key, the bytes of the state in memory
    pub state_size: BTreeMap<u32, usize>,
+    /// For each key, the last execution time of flow in unix timestamp milliseconds.
+    pub last_exec_time_map: BTreeMap<FlowId, i64>,
 }

 impl From<FlowStateValue> for FlowStat {
    fn from(value: FlowStateValue) -> Self {
        Self {
            state_size: value.state_size,
+            last_exec_time_map: value.last_exec_time_map,
        }
    }
 }
@@ -157,6 +168,7 @@ impl From<FlowStat> for FlowStateValue {
    fn from(value: FlowStat) -> Self {
        Self {
            state_size: value.state_size,
+            last_exec_time_map: value.last_exec_time_map,
        }
    }
 }
--- a/src/common/meta/src/key/test_utils.rs
+++ b/src/common/meta/src/key/test_utils.rs
@@ -40,7 +40,7 @@ pub fn new_test_table_info_with_name<I: IntoIterator<Item = u32>>(
        .build()
        .unwrap();

-    let meta = TableMetaBuilder::default()
+    let meta = TableMetaBuilder::empty()
        .schema(Arc::new(schema))
        .primary_key_indices(vec![0])
        .engine("engine")
--- a/src/common/meta/src/kv_backend.rs
+++ b/src/common/meta/src/kv_backend.rs
@@ -31,7 +31,7 @@ use crate::rpc::KeyValue;
 pub mod chroot;
 pub mod etcd;
 pub mod memory;
-#[cfg(feature = "pg_kvbackend")]
+#[cfg(any(feature = "mysql_kvbackend", feature = "pg_kvbackend"))]
 pub mod rds;
 pub mod test;
 pub mod txn;
--- a/src/common/meta/src/kv_backend/memory.rs
+++ b/src/common/meta/src/kv_backend/memory.rs
@@ -14,13 +14,11 @@

 use std::any::Any;
 use std::collections::BTreeMap;
-use std::fmt::{Display, Formatter};
 use std::marker::PhantomData;
 use std::sync::{Arc, RwLock};

 use async_trait::async_trait;
 use common_error::ext::ErrorExt;
-use serde::Serializer;

 use super::{KvBackendRef, ResettableKvBackend};
 use crate::kv_backend::txn::{Txn, TxnOp, TxnOpResponse, TxnRequest, TxnResponse};
@@ -38,19 +36,6 @@ pub struct MemoryKvBackend<T> {
    _phantom: PhantomData<T>,
 }

-impl<T> Display for MemoryKvBackend<T> {
-    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
-        let kvs = self.kvs.read().unwrap();
-        for (k, v) in kvs.iter() {
-            f.serialize_str(&String::from_utf8_lossy(k))?;
-            f.serialize_str(" -> ")?;
-            f.serialize_str(&String::from_utf8_lossy(v))?;
-            f.serialize_str("\n")?;
-        }
-        Ok(())
-    }
-}
-
 impl<T> Default for MemoryKvBackend<T> {
    fn default() -> Self {
        Self {
--- a/src/common/meta/src/kv_backend/rds.rs
+++ b/src/common/meta/src/kv_backend/rds.rs
@@ -33,10 +33,16 @@ use crate::rpc::store::{
 };
 use crate::rpc::KeyValue;

+#[cfg(feature = "pg_kvbackend")]
 mod postgres;
-
+#[cfg(feature = "pg_kvbackend")]
 pub use postgres::PgStore;

+#[cfg(feature = "mysql_kvbackend")]
+mod mysql;
+#[cfg(feature = "mysql_kvbackend")]
+pub use mysql::MySqlStore;
+
 const RDS_STORE_TXN_RETRY_COUNT: usize = 3;

 /// Query executor for rds. It can execute queries or generate a transaction executor.
@@ -106,6 +112,14 @@ impl<T: Executor> ExecutorImpl<'_, T> {
        }
    }

+    #[warn(dead_code)] // Used in #[cfg(feature = "mysql_kvbackend")]
+    async fn execute(&mut self, query: &str, params: &Vec<&Vec<u8>>) -> Result<()> {
+        match self {
+            Self::Default(executor) => executor.execute(query, params).await,
+            Self::Txn(executor) => executor.execute(query, params).await,
+        }
+    }
+
    async fn commit(self) -> Result<()> {
        match self {
            Self::Txn(executor) => executor.commit().await,
--- a/src/common/meta/src/kv_backend/rds/mysql.rs
+++ b/src/common/meta/src/kv_backend/rds/mysql.rs
@@ -0,0 +1,650 @@
+// 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::marker::PhantomData;
+use std::sync::Arc;
+
+use common_telemetry::debug;
+use snafu::ResultExt;
+use sqlx::mysql::MySqlRow;
+use sqlx::pool::Pool;
+use sqlx::{MySql, MySqlPool, Row, Transaction as MySqlTransaction};
+
+use crate::error::{CreateMySqlPoolSnafu, MySqlExecutionSnafu, MySqlTransactionSnafu, Result};
+use crate::kv_backend::rds::{
+    Executor, ExecutorFactory, ExecutorImpl, KvQueryExecutor, RdsStore, Transaction,
+    RDS_STORE_TXN_RETRY_COUNT,
+};
+use crate::kv_backend::KvBackendRef;
+use crate::rpc::store::{
+    BatchDeleteRequest, BatchDeleteResponse, BatchGetRequest, BatchGetResponse, BatchPutRequest,
+    BatchPutResponse, DeleteRangeRequest, DeleteRangeResponse, RangeRequest, RangeResponse,
+};
+use crate::rpc::KeyValue;
+
+type MySqlClient = Arc<Pool<MySql>>;
+pub struct MySqlTxnClient(MySqlTransaction<'static, MySql>);
+
+fn key_value_from_row(row: MySqlRow) -> KeyValue {
+    // Safety: key and value are the first two columns in the row
+    KeyValue {
+        key: row.get_unchecked(0),
+        value: row.get_unchecked(1),
+    }
+}
+
+const EMPTY: &[u8] = &[0];
+
+/// Type of range template.
+#[derive(Debug, Clone, Copy)]
+enum RangeTemplateType {
+    Point,
+    Range,
+    Full,
+    LeftBounded,
+    Prefix,
+}
+
+/// Builds params for the given range template type.
+impl RangeTemplateType {
+    fn build_params(&self, mut key: Vec<u8>, range_end: Vec<u8>) -> Vec<Vec<u8>> {
+        match self {
+            RangeTemplateType::Point => vec![key],
+            RangeTemplateType::Range => vec![key, range_end],
+            RangeTemplateType::Full => vec![],
+            RangeTemplateType::LeftBounded => vec![key],
+            RangeTemplateType::Prefix => {
+                key.push(b'%');
+                vec![key]
+            }
+        }
+    }
+}
+
+/// Templates for range request.
+#[derive(Debug, Clone)]
+struct RangeTemplate {
+    point: String,
+    range: String,
+    full: String,
+    left_bounded: String,
+    prefix: String,
+}
+
+impl RangeTemplate {
+    /// Gets the template for the given type.
+    fn get(&self, typ: RangeTemplateType) -> &str {
+        match typ {
+            RangeTemplateType::Point => &self.point,
+            RangeTemplateType::Range => &self.range,
+            RangeTemplateType::Full => &self.full,
+            RangeTemplateType::LeftBounded => &self.left_bounded,
+            RangeTemplateType::Prefix => &self.prefix,
+        }
+    }
+
+    /// Adds limit to the template.
+    fn with_limit(template: &str, limit: i64) -> String {
+        if limit == 0 {
+            return format!("{};", template);
+        }
+        format!("{} LIMIT {};", template, limit)
+    }
+}
+
+fn is_prefix_range(start: &[u8], end: &[u8]) -> bool {
+    if start.len() != end.len() {
+        return false;
+    }
+    let l = start.len();
+    let same_prefix = start[0..l - 1] == end[0..l - 1];
+    if let (Some(rhs), Some(lhs)) = (start.last(), end.last()) {
+        return same_prefix && (*rhs + 1) == *lhs;
+    }
+    false
+}
+
+/// Determine the template type for range request.
+fn range_template(key: &[u8], range_end: &[u8]) -> RangeTemplateType {
+    match (key, range_end) {
+        (_, &[]) => RangeTemplateType::Point,
+        (EMPTY, EMPTY) => RangeTemplateType::Full,
+        (_, EMPTY) => RangeTemplateType::LeftBounded,
+        (start, end) => {
+            if is_prefix_range(start, end) {
+                RangeTemplateType::Prefix
+            } else {
+                RangeTemplateType::Range
+            }
+        }
+    }
+}
+
+/// Generate in placeholders for MySQL.
+fn mysql_generate_in_placeholders(from: usize, to: usize) -> Vec<String> {
+    (from..=to).map(|_| "?".to_string()).collect()
+}
+
+/// Factory for building sql templates.
+struct MySqlTemplateFactory<'a> {
+    table_name: &'a str,
+}
+
+impl<'a> MySqlTemplateFactory<'a> {
+    /// Creates a new [`SqlTemplateFactory`] with the given table name.
+    fn new(table_name: &'a str) -> Self {
+        Self { table_name }
+    }
+
+    /// Builds the template set for the given table name.
+    fn build(&self) -> MySqlTemplateSet {
+        let table_name = self.table_name;
+        // Some of queries don't end with `;`, because we need to add `LIMIT` clause.
+        MySqlTemplateSet {
+            table_name: table_name.to_string(),
+            create_table_statement: format!(
+                // Cannot be more than 3072 bytes in PRIMARY KEY
+                "CREATE TABLE IF NOT EXISTS {table_name}(k VARBINARY(3072) PRIMARY KEY, v BLOB);",
+            ),
+            range_template: RangeTemplate {
+                point: format!("SELECT k, v FROM {table_name} WHERE k = ?"),
+                range: format!("SELECT k, v FROM {table_name} WHERE k >= ? AND k < ? ORDER BY k"),
+                full: format!("SELECT k, v FROM {table_name} ? ORDER BY k"),
+                left_bounded: format!("SELECT k, v FROM {table_name} WHERE k >= ? ORDER BY k"),
+                prefix: format!("SELECT k, v FROM {table_name} WHERE k LIKE ? ORDER BY k"),
+            },
+            delete_template: RangeTemplate {
+                point: format!("DELETE FROM {table_name} WHERE k = ?;"),
+                range: format!("DELETE FROM {table_name} WHERE k >= ? AND k < ?;"),
+                full: format!("DELETE FROM {table_name}"),
+                left_bounded: format!("DELETE FROM {table_name} WHERE k >= ?;"),
+                prefix: format!("DELETE FROM {table_name} WHERE k LIKE ?;"),
+            },
+        }
+    }
+}
+
+/// Templates for the given table name.
+#[derive(Debug, Clone)]
+pub struct MySqlTemplateSet {
+    table_name: String,
+    create_table_statement: String,
+    range_template: RangeTemplate,
+    delete_template: RangeTemplate,
+}
+
+impl MySqlTemplateSet {
+    /// Generates the sql for batch get.
+    fn generate_batch_get_query(&self, key_len: usize) -> String {
+        let table_name = &self.table_name;
+        let in_clause = mysql_generate_in_placeholders(1, key_len).join(", ");
+        format!("SELECT k, v FROM {table_name} WHERE k in ({});", in_clause)
+    }
+
+    /// Generates the sql for batch delete.
+    fn generate_batch_delete_query(&self, key_len: usize) -> String {
+        let table_name = &self.table_name;
+        let in_clause = mysql_generate_in_placeholders(1, key_len).join(", ");
+        format!("DELETE FROM {table_name} WHERE k in ({});", in_clause)
+    }
+
+    /// Generates the sql for batch upsert.
+    /// For MySQL, it also generates a select query to get the previous values.
+    fn generate_batch_upsert_query(&self, kv_len: usize) -> (String, String) {
+        let table_name = &self.table_name;
+        let in_placeholders: Vec<String> = (1..=kv_len).map(|_| "?".to_string()).collect();
+        let in_clause = in_placeholders.join(", ");
+        let mut values_placeholders = Vec::new();
+        for _ in 0..kv_len {
+            values_placeholders.push("(?, ?)".to_string());
+        }
+        let values_clause = values_placeholders.join(", ");
+
+        (
+            format!(r#"SELECT k, v FROM {table_name} WHERE k IN ({in_clause})"#,),
+            format!(
+                r#"INSERT INTO {table_name} (k, v) VALUES {values_clause} ON DUPLICATE KEY UPDATE v = VALUES(v);"#,
+            ),
+        )
+    }
+}
+
+#[async_trait::async_trait]
+impl Executor for MySqlClient {
+    type Transaction<'a>
+        = MySqlTxnClient
+    where
+        Self: 'a;
+
+    fn name() -> &'static str {
+        "MySql"
+    }
+
+    async fn query(&mut self, raw_query: &str, params: &[&Vec<u8>]) -> Result<Vec<KeyValue>> {
+        let query = sqlx::query(raw_query);
+        let query = params.iter().fold(query, |query, param| query.bind(param));
+        let rows = query
+            .fetch_all(&**self)
+            .await
+            .context(MySqlExecutionSnafu { sql: raw_query })?;
+        Ok(rows.into_iter().map(key_value_from_row).collect())
+    }
+
+    async fn execute(&mut self, raw_query: &str, params: &[&Vec<u8>]) -> Result<()> {
+        let query = sqlx::query(raw_query);
+        let query = params.iter().fold(query, |query, param| query.bind(param));
+        query
+            .execute(&**self)
+            .await
+            .context(MySqlExecutionSnafu { sql: raw_query })?;
+        Ok(())
+    }
+
+    async fn txn_executor<'a>(&'a mut self) -> Result<Self::Transaction<'a>> {
+        // sqlx has no isolation level support for now, so we have to set it manually.
+        // TODO(CookiePie): Waiting for https://github.com/launchbadge/sqlx/pull/3614 and remove this.
+        sqlx::query("SET SESSION TRANSACTION ISOLATION LEVEL SERIALIZABLE")
+            .execute(&**self)
+            .await
+            .context(MySqlExecutionSnafu {
+                sql: "SET SESSION TRANSACTION ISOLATION LEVEL SERIALIZABLE",
+            })?;
+        let txn = self
+            .begin()
+            .await
+            .context(MySqlExecutionSnafu { sql: "begin" })?;
+        Ok(MySqlTxnClient(txn))
+    }
+}
+
+#[async_trait::async_trait]
+impl Transaction<'_> for MySqlTxnClient {
+    async fn query(&mut self, raw_query: &str, params: &[&Vec<u8>]) -> Result<Vec<KeyValue>> {
+        let query = sqlx::query(raw_query);
+        let query = params.iter().fold(query, |query, param| query.bind(param));
+        // As said in https://docs.rs/sqlx/latest/sqlx/trait.Executor.html, we need a `&mut *transaction`. Weird.
+        let rows = query
+            .fetch_all(&mut *(self.0))
+            .await
+            .context(MySqlExecutionSnafu { sql: raw_query })?;
+        Ok(rows.into_iter().map(key_value_from_row).collect())
+    }
+
+    async fn execute(&mut self, raw_query: &str, params: &[&Vec<u8>]) -> Result<()> {
+        let query = sqlx::query(raw_query);
+        let query = params.iter().fold(query, |query, param| query.bind(param));
+        // As said in https://docs.rs/sqlx/latest/sqlx/trait.Executor.html, we need a `&mut *transaction`. Weird.
+        query
+            .execute(&mut *(self.0))
+            .await
+            .context(MySqlExecutionSnafu { sql: raw_query })?;
+        Ok(())
+    }
+
+    /// Caution: sqlx will stuck on the query if two transactions conflict with each other.
+    /// Don't know if it's a feature or it depends on the database. Be careful.
+    async fn commit(self) -> Result<()> {
+        self.0.commit().await.context(MySqlTransactionSnafu {
+            operation: "commit",
+        })?;
+        Ok(())
+    }
+}
+
+pub struct MySqlExecutorFactory {
+    pool: Arc<Pool<MySql>>,
+}
+
+#[async_trait::async_trait]
+impl ExecutorFactory<MySqlClient> for MySqlExecutorFactory {
+    async fn default_executor(&self) -> Result<MySqlClient> {
+        Ok(self.pool.clone())
+    }
+
+    async fn txn_executor<'a>(
+        &self,
+        default_executor: &'a mut MySqlClient,
+    ) -> Result<MySqlTxnClient> {
+        default_executor.txn_executor().await
+    }
+}
+
+/// A MySQL-backed key-value store.
+/// It uses [sqlx::Pool<MySql>] as the connection pool for [RdsStore].
+pub type MySqlStore = RdsStore<MySqlClient, MySqlExecutorFactory, MySqlTemplateSet>;
+
+#[async_trait::async_trait]
+impl KvQueryExecutor<MySqlClient> for MySqlStore {
+    async fn range_with_query_executor(
+        &self,
+        query_executor: &mut ExecutorImpl<'_, MySqlClient>,
+        req: RangeRequest,
+    ) -> Result<RangeResponse> {
+        let template_type = range_template(&req.key, &req.range_end);
+        let template = self.sql_template_set.range_template.get(template_type);
+        let params = template_type.build_params(req.key, req.range_end);
+        let params_ref = params.iter().collect::<Vec<_>>();
+        // Always add 1 to limit to check if there is more data
+        let query =
+            RangeTemplate::with_limit(template, if req.limit == 0 { 0 } else { req.limit + 1 });
+        let limit = req.limit as usize;
+        debug!("query: {:?}, params: {:?}", query, params);
+        let mut kvs = query_executor.query(&query, &params_ref).await?;
+        if req.keys_only {
+            kvs.iter_mut().for_each(|kv| kv.value = vec![]);
+        }
+        // If limit is 0, we always return all data
+        if limit == 0 || kvs.len() <= limit {
+            return Ok(RangeResponse { kvs, more: false });
+        }
+        // If limit is greater than the number of rows, we remove the last row and set more to true
+        let removed = kvs.pop();
+        debug_assert!(removed.is_some());
+        Ok(RangeResponse { kvs, more: true })
+    }
+
+    async fn batch_put_with_query_executor(
+        &self,
+        query_executor: &mut ExecutorImpl<'_, MySqlClient>,
+        req: BatchPutRequest,
+    ) -> Result<BatchPutResponse> {
+        let mut in_params = Vec::with_capacity(req.kvs.len() * 3);
+        let mut values_params = Vec::with_capacity(req.kvs.len() * 2);
+
+        for kv in &req.kvs {
+            let processed_key = &kv.key;
+            in_params.push(processed_key);
+
+            let processed_value = &kv.value;
+            values_params.push(processed_key);
+            values_params.push(processed_value);
+        }
+        let in_params = in_params.iter().map(|x| x as _).collect::<Vec<_>>();
+        let values_params = values_params.iter().map(|x| x as _).collect::<Vec<_>>();
+        let (select, update) = self
+            .sql_template_set
+            .generate_batch_upsert_query(req.kvs.len());
+
+        // Fast path: if we don't need previous kvs, we can just upsert the keys.
+        if !req.prev_kv {
+            query_executor.execute(&update, &values_params).await?;
+            return Ok(BatchPutResponse::default());
+        }
+        // Should use transaction to ensure atomicity.
+        if let ExecutorImpl::Default(query_executor) = query_executor {
+            let txn = query_executor.txn_executor().await?;
+            let mut txn = ExecutorImpl::Txn(txn);
+            let res = self.batch_put_with_query_executor(&mut txn, req).await;
+            txn.commit().await?;
+            return res;
+        }
+        let prev_kvs = query_executor.query(&select, &in_params).await?;
+        query_executor.execute(&update, &values_params).await?;
+        Ok(BatchPutResponse { prev_kvs })
+    }
+
+    async fn batch_get_with_query_executor(
+        &self,
+        query_executor: &mut ExecutorImpl<'_, MySqlClient>,
+        req: BatchGetRequest,
+    ) -> Result<BatchGetResponse> {
+        if req.keys.is_empty() {
+            return Ok(BatchGetResponse { kvs: vec![] });
+        }
+        let query = self
+            .sql_template_set
+            .generate_batch_get_query(req.keys.len());
+        let params = req.keys.iter().map(|x| x as _).collect::<Vec<_>>();
+        let kvs = query_executor.query(&query, &params).await?;
+        Ok(BatchGetResponse { kvs })
+    }
+
+    async fn delete_range_with_query_executor(
+        &self,
+        query_executor: &mut ExecutorImpl<'_, MySqlClient>,
+        req: DeleteRangeRequest,
+    ) -> Result<DeleteRangeResponse> {
+        // Since we need to know the number of deleted keys, we have no fast path here.
+        // Should use transaction to ensure atomicity.
+        if let ExecutorImpl::Default(query_executor) = query_executor {
+            let txn = query_executor.txn_executor().await?;
+            let mut txn = ExecutorImpl::Txn(txn);
+            let res = self.delete_range_with_query_executor(&mut txn, req).await;
+            txn.commit().await?;
+            return res;
+        }
+        let range_get_req = RangeRequest {
+            key: req.key.clone(),
+            range_end: req.range_end.clone(),
+            limit: 0,
+            keys_only: false,
+        };
+        let prev_kvs = self
+            .range_with_query_executor(query_executor, range_get_req)
+            .await?
+            .kvs;
+        let template_type = range_template(&req.key, &req.range_end);
+        let template = self.sql_template_set.delete_template.get(template_type);
+        let params = template_type.build_params(req.key, req.range_end);
+        let params_ref = params.iter().map(|x| x as _).collect::<Vec<_>>();
+        query_executor.execute(template, &params_ref).await?;
+        let mut resp = DeleteRangeResponse::new(prev_kvs.len() as i64);
+        if req.prev_kv {
+            resp.with_prev_kvs(prev_kvs);
+        }
+        Ok(resp)
+    }
+
+    async fn batch_delete_with_query_executor(
+        &self,
+        query_executor: &mut ExecutorImpl<'_, MySqlClient>,
+        req: BatchDeleteRequest,
+    ) -> Result<BatchDeleteResponse> {
+        if req.keys.is_empty() {
+            return Ok(BatchDeleteResponse::default());
+        }
+        let query = self
+            .sql_template_set
+            .generate_batch_delete_query(req.keys.len());
+        let params = req.keys.iter().map(|x| x as _).collect::<Vec<_>>();
+        // Fast path: if we don't need previous kvs, we can just delete the keys.
+        if !req.prev_kv {
+            query_executor.execute(&query, &params).await?;
+            return Ok(BatchDeleteResponse::default());
+        }
+        // Should use transaction to ensure atomicity.
+        if let ExecutorImpl::Default(query_executor) = query_executor {
+            let txn = query_executor.txn_executor().await?;
+            let mut txn = ExecutorImpl::Txn(txn);
+            let res = self.batch_delete_with_query_executor(&mut txn, req).await;
+            txn.commit().await?;
+            return res;
+        }
+        // Should get previous kvs first
+        let batch_get_req = BatchGetRequest {
+            keys: req.keys.clone(),
+        };
+        let prev_kvs = self
+            .batch_get_with_query_executor(query_executor, batch_get_req)
+            .await?
+            .kvs;
+        // Pure `DELETE` has no return value, so we need to use `execute` instead of `query`.
+        query_executor.execute(&query, &params).await?;
+        if req.prev_kv {
+            Ok(BatchDeleteResponse { prev_kvs })
+        } else {
+            Ok(BatchDeleteResponse::default())
+        }
+    }
+}
+
+impl MySqlStore {
+    /// Create [MySqlStore] impl of [KvBackendRef] from url.
+    pub async fn with_url(url: &str, table_name: &str, max_txn_ops: usize) -> Result<KvBackendRef> {
+        let pool = MySqlPool::connect(url)
+            .await
+            .context(CreateMySqlPoolSnafu)?;
+        Self::with_mysql_pool(pool, table_name, max_txn_ops).await
+    }
+
+    /// Create [MySqlStore] impl of [KvBackendRef] from [sqlx::Pool<MySql>].
+    pub async fn with_mysql_pool(
+        pool: Pool<MySql>,
+        table_name: &str,
+        max_txn_ops: usize,
+    ) -> Result<KvBackendRef> {
+        // This step ensures the mysql metadata backend is ready to use.
+        // We check if greptime_metakv table exists, and we will create a new table
+        // if it does not exist.
+        let sql_template_set = MySqlTemplateFactory::new(table_name).build();
+        sqlx::query(&sql_template_set.create_table_statement)
+            .execute(&pool)
+            .await
+            .context(MySqlExecutionSnafu {
+                sql: sql_template_set.create_table_statement.to_string(),
+            })?;
+        Ok(Arc::new(MySqlStore {
+            max_txn_ops,
+            sql_template_set,
+            txn_retry_count: RDS_STORE_TXN_RETRY_COUNT,
+            executor_factory: MySqlExecutorFactory {
+                pool: Arc::new(pool),
+            },
+            _phantom: PhantomData,
+        }))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use common_telemetry::init_default_ut_logging;
+
+    use super::*;
+    use crate::kv_backend::test::{
+        prepare_kv_with_prefix, test_kv_batch_delete_with_prefix, test_kv_batch_get_with_prefix,
+        test_kv_compare_and_put_with_prefix, test_kv_delete_range_with_prefix,
+        test_kv_put_with_prefix, test_kv_range_2_with_prefix, test_kv_range_with_prefix,
+        test_txn_compare_equal, test_txn_compare_greater, test_txn_compare_less,
+        test_txn_compare_not_equal, test_txn_one_compare_op, text_txn_multi_compare_op,
+        unprepare_kv,
+    };
+
+    async fn build_mysql_kv_backend(table_name: &str) -> Option<MySqlStore> {
+        init_default_ut_logging();
+        let endpoints = std::env::var("GT_MYSQL_ENDPOINTS").unwrap_or_default();
+        if endpoints.is_empty() {
+            return None;
+        }
+        let pool = MySqlPool::connect(&endpoints).await.unwrap();
+        let sql_templates = MySqlTemplateFactory::new(table_name).build();
+        sqlx::query(&sql_templates.create_table_statement)
+            .execute(&pool)
+            .await
+            .unwrap();
+        Some(MySqlStore {
+            max_txn_ops: 128,
+            sql_template_set: sql_templates,
+            txn_retry_count: RDS_STORE_TXN_RETRY_COUNT,
+            executor_factory: MySqlExecutorFactory {
+                pool: Arc::new(pool),
+            },
+            _phantom: PhantomData,
+        })
+    }
+
+    #[tokio::test]
+    async fn test_mysql_put() {
+        let kv_backend = build_mysql_kv_backend("put_test").await.unwrap();
+        let prefix = b"put/";
+        prepare_kv_with_prefix(&kv_backend, prefix.to_vec()).await;
+        test_kv_put_with_prefix(&kv_backend, prefix.to_vec()).await;
+        unprepare_kv(&kv_backend, prefix).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_range() {
+        let kv_backend = build_mysql_kv_backend("range_test").await.unwrap();
+        let prefix = b"range/";
+        prepare_kv_with_prefix(&kv_backend, prefix.to_vec()).await;
+        test_kv_range_with_prefix(&kv_backend, prefix.to_vec()).await;
+        unprepare_kv(&kv_backend, prefix).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_range_2() {
+        let kv_backend = build_mysql_kv_backend("range2_test").await.unwrap();
+        let prefix = b"range2/";
+        test_kv_range_2_with_prefix(&kv_backend, prefix.to_vec()).await;
+        unprepare_kv(&kv_backend, prefix).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_batch_get() {
+        let kv_backend = build_mysql_kv_backend("batch_get_test").await.unwrap();
+        let prefix = b"batch_get/";
+        prepare_kv_with_prefix(&kv_backend, prefix.to_vec()).await;
+        test_kv_batch_get_with_prefix(&kv_backend, prefix.to_vec()).await;
+        unprepare_kv(&kv_backend, prefix).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_batch_delete() {
+        let kv_backend = build_mysql_kv_backend("batch_delete_test").await.unwrap();
+        let prefix = b"batch_delete/";
+        prepare_kv_with_prefix(&kv_backend, prefix.to_vec()).await;
+        test_kv_delete_range_with_prefix(&kv_backend, prefix.to_vec()).await;
+        unprepare_kv(&kv_backend, prefix).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_batch_delete_with_prefix() {
+        let kv_backend = build_mysql_kv_backend("batch_delete_with_prefix_test")
+            .await
+            .unwrap();
+        let prefix = b"batch_delete/";
+        prepare_kv_with_prefix(&kv_backend, prefix.to_vec()).await;
+        test_kv_batch_delete_with_prefix(&kv_backend, prefix.to_vec()).await;
+        unprepare_kv(&kv_backend, prefix).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_delete_range() {
+        let kv_backend = build_mysql_kv_backend("delete_range_test").await.unwrap();
+        let prefix = b"delete_range/";
+        prepare_kv_with_prefix(&kv_backend, prefix.to_vec()).await;
+        test_kv_delete_range_with_prefix(&kv_backend, prefix.to_vec()).await;
+        unprepare_kv(&kv_backend, prefix).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_compare_and_put() {
+        let kv_backend = build_mysql_kv_backend("compare_and_put_test")
+            .await
+            .unwrap();
+        let prefix = b"compare_and_put/";
+        let kv_backend = Arc::new(kv_backend);
+        test_kv_compare_and_put_with_prefix(kv_backend.clone(), prefix.to_vec()).await;
+    }
+
+    #[tokio::test]
+    async fn test_mysql_txn() {
+        let kv_backend = build_mysql_kv_backend("txn_test").await.unwrap();
+        test_txn_one_compare_op(&kv_backend).await;
+        text_txn_multi_compare_op(&kv_backend).await;
+        test_txn_compare_equal(&kv_backend).await;
+        test_txn_compare_greater(&kv_backend).await;
+        test_txn_compare_less(&kv_backend).await;
+        test_txn_compare_not_equal(&kv_backend).await;
+    }
+}
--- a/src/common/meta/src/kv_backend/rds/postgres.rs
+++ b/src/common/meta/src/kv_backend/rds/postgres.rs
@@ -153,6 +153,7 @@ impl<'a> PgSqlTemplateFactory<'a> {
    /// Builds the template set for the given table name.
    fn build(&self) -> PgSqlTemplateSet {
        let table_name = self.table_name;
+        // Some of queries don't end with `;`, because we need to add `LIMIT` clause.
        PgSqlTemplateSet {
            table_name: table_name.to_string(),
            create_table_statement: format!(
--- a/src/common/meta/src/rpc/procedure.rs
+++ b/src/common/meta/src/rpc/procedure.rs
@@ -34,6 +34,24 @@ pub struct MigrateRegionRequest {
    pub timeout: Duration,
 }

+/// A request to add region follower.
+#[derive(Debug, Clone)]
+pub struct AddRegionFollowerRequest {
+    /// The region id to add follower.
+    pub region_id: u64,
+    /// The peer id to add follower.
+    pub peer_id: u64,
+}
+
+/// A request to remove region follower.
+#[derive(Debug, Clone)]
+pub struct RemoveRegionFollowerRequest {
+    /// The region id to remove follower.
+    pub region_id: u64,
+    /// The peer id to remove follower.
+    pub peer_id: u64,
+}
+
 /// Cast the protobuf [`ProcedureId`] to common [`ProcedureId`].
 pub fn pb_pid_to_pid(pid: &PbProcedureId) -> Result<ProcedureId> {
    ProcedureId::parse_str(&String::from_utf8_lossy(&pid.key)).with_context(|_| {
--- a/src/common/recordbatch/src/filter.rs
+++ b/src/common/recordbatch/src/filter.rs
@@ -26,6 +26,7 @@ use datafusion_common::cast::{as_boolean_array, as_null_array};
 use datafusion_common::{internal_err, DataFusionError, ScalarValue};
 use datatypes::arrow::array::{Array, BooleanArray, RecordBatch};
 use datatypes::arrow::compute::filter_record_batch;
+use datatypes::compute::or_kleene;
 use datatypes::vectors::VectorRef;
 use snafu::ResultExt;

@@ -47,6 +48,8 @@ pub struct SimpleFilterEvaluator {
    literal: Scalar<ArrayRef>,
    /// The operator.
    op: Operator,
+    /// Only used when the operator is `Or`-chain.
+    literal_list: Vec<Scalar<ArrayRef>>,
 }

 impl SimpleFilterEvaluator {
@@ -69,6 +72,7 @@ impl SimpleFilterEvaluator {
            column_name,
            literal: val.to_scalar().ok()?,
            op,
+            literal_list: vec![],
        })
    }

@@ -83,6 +87,35 @@ impl SimpleFilterEvaluator {
                    | Operator::LtEq
                    | Operator::Gt
                    | Operator::GtEq => {}
+                    Operator::Or => {
+                        let lhs = Self::try_new(&binary.left)?;
+                        let rhs = Self::try_new(&binary.right)?;
+                        if lhs.column_name != rhs.column_name
+                            || !matches!(lhs.op, Operator::Eq | Operator::Or)
+                            || !matches!(rhs.op, Operator::Eq | Operator::Or)
+                        {
+                            return None;
+                        }
+                        let mut list = vec![];
+                        let placeholder_literal = lhs.literal.clone();
+                        // above check guarantees the op is either `Eq` or `Or`
+                        if matches!(lhs.op, Operator::Or) {
+                            list.extend(lhs.literal_list);
+                        } else {
+                            list.push(lhs.literal);
+                        }
+                        if matches!(rhs.op, Operator::Or) {
+                            list.extend(rhs.literal_list);
+                        } else {
+                            list.push(rhs.literal);
+                        }
+                        return Some(Self {
+                            column_name: lhs.column_name,
+                            literal: placeholder_literal,
+                            op: Operator::Or,
+                            literal_list: list,
+                        });
+                    }
                    _ => return None,
                }

@@ -103,6 +136,7 @@ impl SimpleFilterEvaluator {
                    column_name: lhs.name.clone(),
                    literal,
                    op,
+                    literal_list: vec![],
                })
            }
            _ => None,
@@ -118,19 +152,19 @@ impl SimpleFilterEvaluator {
        let input = input
            .to_scalar()
            .with_context(|_| ToArrowScalarSnafu { v: input.clone() })?;
-        let result = self.evaluate_datum(&input)?;
+        let result = self.evaluate_datum(&input, 1)?;
        Ok(result.value(0))
    }

    pub fn evaluate_array(&self, input: &ArrayRef) -> Result<BooleanBuffer> {
-        self.evaluate_datum(input)
+        self.evaluate_datum(input, input.len())
    }

    pub fn evaluate_vector(&self, input: &VectorRef) -> Result<BooleanBuffer> {
-        self.evaluate_datum(&input.to_arrow_array())
+        self.evaluate_datum(&input.to_arrow_array(), input.len())
    }

-    fn evaluate_datum(&self, input: &impl Datum) -> Result<BooleanBuffer> {
+    fn evaluate_datum(&self, input: &impl Datum, input_len: usize) -> Result<BooleanBuffer> {
        let result = match self.op {
            Operator::Eq => cmp::eq(input, &self.literal),
            Operator::NotEq => cmp::neq(input, &self.literal),
@@ -138,6 +172,15 @@ impl SimpleFilterEvaluator {
            Operator::LtEq => cmp::lt_eq(input, &self.literal),
            Operator::Gt => cmp::gt(input, &self.literal),
            Operator::GtEq => cmp::gt_eq(input, &self.literal),
+            Operator::Or => {
+                // OR operator stands for OR-chained EQs (or INLIST in other words)
+                let mut result: BooleanArray = vec![false; input_len].into();
+                for literal in &self.literal_list {
+                    let rhs = cmp::eq(input, literal).context(ArrowComputeSnafu)?;
+                    result = or_kleene(&result, &rhs).context(ArrowComputeSnafu)?;
+                }
+                Ok(result)
+            }
            _ => {
                return UnsupportedOperationSnafu {
                    reason: format!("{:?}", self.op),
@@ -349,4 +392,49 @@ mod test {
        let expected = datatypes::arrow::array::Int32Array::from(vec![5, 6]);
        assert_eq!(first_column_values, &expected);
    }
+
+    #[test]
+    fn test_complex_filter_expression() {
+        // Create an expression tree for: col = 'B' OR col = 'C' OR col = 'D'
+        let col_eq_b = col("col").eq(lit("B"));
+        let col_eq_c = col("col").eq(lit("C"));
+        let col_eq_d = col("col").eq(lit("D"));
+
+        // Build the OR chain
+        let col_or_expr = col_eq_b.or(col_eq_c).or(col_eq_d);
+
+        // Check that SimpleFilterEvaluator can handle OR chain
+        let or_evaluator = SimpleFilterEvaluator::try_new(&col_or_expr).unwrap();
+        assert_eq!(or_evaluator.column_name, "col");
+        assert_eq!(or_evaluator.op, Operator::Or);
+        assert_eq!(or_evaluator.literal_list.len(), 3);
+        assert_eq!(format!("{:?}", or_evaluator.literal_list), "[Scalar(StringArray\n[\n  \"B\",\n]), Scalar(StringArray\n[\n  \"C\",\n]), Scalar(StringArray\n[\n  \"D\",\n])]");
+
+        // Create a schema and batch for testing
+        let schema = Schema::new(vec![Field::new("col", DataType::Utf8, false)]);
+        let df_schema = DFSchema::try_from(schema.clone()).unwrap();
+        let props = ExecutionProps::new();
+        let physical_expr = create_physical_expr(&col_or_expr, &df_schema, &props).unwrap();
+
+        // Create test data
+        let col_data = Arc::new(datatypes::arrow::array::StringArray::from(vec![
+            "B", "C", "E", "B", "C", "D", "F",
+        ]));
+        let batch = RecordBatch::try_new(Arc::new(schema), vec![col_data]).unwrap();
+        let expected = datatypes::arrow::array::StringArray::from(vec!["B", "C", "B", "C", "D"]);
+
+        // Filter the batch
+        let filtered_batch = batch_filter(&batch, &physical_expr).unwrap();
+
+        // Expected: rows with col in ("B", "C", "D")
+        // That would be rows 0, 1, 3, 4, 5
+        assert_eq!(filtered_batch.num_rows(), 5);
+
+        let col_filtered = filtered_batch
+            .column(0)
+            .as_any()
+            .downcast_ref::<datatypes::arrow::array::StringArray>()
+            .unwrap();
+        assert_eq!(col_filtered, &expected);
+    }
 }
--- a/src/common/telemetry/src/logging.rs
+++ b/src/common/telemetry/src/logging.rs
@@ -111,7 +111,7 @@ impl Eq for LoggingOptions {}
 impl Default for LoggingOptions {
    fn default() -> Self {
        Self {
-            dir: "/tmp/greptimedb/logs".to_string(),
+            dir: "./greptimedb_data/logs".to_string(),
            level: None,
            log_format: LogFormat::Text,
            enable_otlp_tracing: false,
--- a/src/common/time/src/datetime.rs
+++ b/src/common/time/src/datetime.rs
@@ -1,407 +0,0 @@
-// 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, Formatter, Write};
-
-use chrono::{
-    Days, LocalResult, Months, NaiveDateTime, TimeDelta, TimeZone as ChronoTimeZone, Utc,
-};
-use serde::{Deserialize, Serialize};
-use snafu::ResultExt;
-
-use crate::error::{InvalidDateStrSnafu, Result};
-use crate::interval::{IntervalDayTime, IntervalMonthDayNano, IntervalYearMonth};
-use crate::timezone::{get_timezone, Timezone};
-use crate::util::{datetime_to_utc, format_utc_datetime};
-use crate::Date;
-
-const DATETIME_FORMAT: &str = "%F %H:%M:%S%.f";
-const DATETIME_FORMAT_WITH_TZ: &str = "%F %H:%M:%S%.f%z";
-
-/// [DateTime] represents the **milliseconds elapsed since "1970-01-01 00:00:00 UTC" (UNIX Epoch)**.
-#[derive(
-    Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, Default, Serialize, Deserialize,
-)]
-pub struct DateTime(i64);
-
-impl Display for DateTime {
-    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
-        if let Some(abs_time) = chrono::DateTime::from_timestamp_millis(self.0) {
-            write!(
-                f,
-                "{}",
-                format_utc_datetime(&abs_time.naive_utc(), DATETIME_FORMAT_WITH_TZ)
-            )
-        } else {
-            write!(f, "DateTime({})", self.0)
-        }
-    }
-}
-
-impl From<DateTime> for serde_json::Value {
-    fn from(d: DateTime) -> Self {
-        serde_json::Value::String(d.to_string())
-    }
-}
-
-impl From<NaiveDateTime> for DateTime {
-    fn from(value: NaiveDateTime) -> Self {
-        DateTime::from(value.and_utc().timestamp_millis())
-    }
-}
-
-impl From<i64> for DateTime {
-    fn from(v: i64) -> Self {
-        Self(v)
-    }
-}
-
-impl From<Date> for DateTime {
-    fn from(value: Date) -> Self {
-        // It's safe, i32 * 86400000 won't be overflow
-        Self(value.to_secs() * 1000)
-    }
-}
-
-impl DateTime {
-    /// Try parsing a string into [`DateTime`] with the system timezone.
-    /// See `DateTime::from_str`.
-    pub fn from_str_system(s: &str) -> Result<Self> {
-        Self::from_str(s, None)
-    }
-
-    /// Try parsing a string into [`DateTime`] with the given timezone.
-    /// Supported format:
-    /// - RFC3339 in the naive UTC timezone.
-    /// - `%F %T`  with the given timezone
-    /// - `%F %T%z`  with the timezone in string
-    pub fn from_str(s: &str, timezone: Option<&Timezone>) -> Result<Self> {
-        let s = s.trim();
-        let timestamp_millis = if let Ok(dt) = chrono::DateTime::parse_from_rfc3339(s) {
-            dt.naive_utc().and_utc().timestamp_millis()
-        } else if let Ok(d) = NaiveDateTime::parse_from_str(s, DATETIME_FORMAT) {
-            match datetime_to_utc(&d, get_timezone(timezone)) {
-                LocalResult::None => {
-                    return InvalidDateStrSnafu { raw: s }.fail();
-                }
-                LocalResult::Single(t) | LocalResult::Ambiguous(t, _) => {
-                    t.and_utc().timestamp_millis()
-                }
-            }
-        } else if let Ok(v) = chrono::DateTime::parse_from_str(s, DATETIME_FORMAT_WITH_TZ) {
-            v.timestamp_millis()
-        } else {
-            return InvalidDateStrSnafu { raw: s }.fail();
-        };
-
-        Ok(Self(timestamp_millis))
-    }
-
-    /// Create a new [DateTime] from milliseconds elapsed since "1970-01-01 00:00:00 UTC" (UNIX Epoch).
-    pub fn new(millis: i64) -> Self {
-        Self(millis)
-    }
-
-    /// Get the milliseconds elapsed since "1970-01-01 00:00:00 UTC" (UNIX Epoch).
-    pub fn val(&self) -> i64 {
-        self.0
-    }
-
-    /// Convert to [NaiveDateTime].
-    pub fn to_chrono_datetime(&self) -> Option<NaiveDateTime> {
-        chrono::DateTime::from_timestamp_millis(self.0).map(|x| x.naive_utc())
-    }
-
-    /// Format DateTime for given format and timezone.
-    /// If `tz==None`, the server default timezone will used.
-    pub fn as_formatted_string(
-        self,
-        pattern: &str,
-        timezone: Option<&Timezone>,
-    ) -> Result<Option<String>> {
-        if let Some(v) = self.to_chrono_datetime() {
-            let mut formatted = String::new();
-
-            match get_timezone(timezone) {
-                Timezone::Offset(offset) => {
-                    write!(
-                        formatted,
-                        "{}",
-                        offset.from_utc_datetime(&v).format(pattern)
-                    )
-                    .context(crate::error::FormatSnafu { pattern })?;
-                }
-                Timezone::Named(tz) => {
-                    write!(formatted, "{}", tz.from_utc_datetime(&v).format(pattern))
-                        .context(crate::error::FormatSnafu { pattern })?;
-                }
-            }
-
-            return Ok(Some(formatted));
-        }
-
-        Ok(None)
-    }
-
-    pub fn to_chrono_datetime_with_timezone(&self, tz: Option<&Timezone>) -> Option<NaiveDateTime> {
-        let datetime = self.to_chrono_datetime();
-        datetime.map(|v| match tz {
-            Some(Timezone::Offset(offset)) => offset.from_utc_datetime(&v).naive_local(),
-            Some(Timezone::Named(tz)) => tz.from_utc_datetime(&v).naive_local(),
-            None => Utc.from_utc_datetime(&v).naive_local(),
-        })
-    }
-
-    // FIXME(yingwen): remove add/sub intervals later
-    /// Adds given [IntervalYearMonth] to the current datetime.
-    pub fn add_year_month(&self, interval: IntervalYearMonth) -> Option<Self> {
-        let naive_datetime = self.to_chrono_datetime()?;
-
-        naive_datetime
-            .checked_add_months(Months::new(interval.months as u32))
-            .map(Into::into)
-    }
-
-    /// Adds given [IntervalDayTime] to the current datetime.
-    pub fn add_day_time(&self, interval: IntervalDayTime) -> Option<Self> {
-        let naive_datetime = self.to_chrono_datetime()?;
-
-        naive_datetime
-            .checked_add_days(Days::new(interval.days as u64))?
-            .checked_add_signed(TimeDelta::milliseconds(interval.milliseconds as i64))
-            .map(Into::into)
-    }
-
-    /// Adds given [IntervalMonthDayNano] to the current datetime.
-    pub fn add_month_day_nano(&self, interval: IntervalMonthDayNano) -> Option<Self> {
-        let naive_datetime = self.to_chrono_datetime()?;
-
-        naive_datetime
-            .checked_add_months(Months::new(interval.months as u32))?
-            .checked_add_days(Days::new(interval.days as u64))?
-            .checked_add_signed(TimeDelta::nanoseconds(interval.nanoseconds))
-            .map(Into::into)
-    }
-
-    /// Subtracts given [IntervalYearMonth] to the current datetime.
-    pub fn sub_year_month(&self, interval: IntervalYearMonth) -> Option<Self> {
-        let naive_datetime = self.to_chrono_datetime()?;
-
-        naive_datetime
-            .checked_sub_months(Months::new(interval.months as u32))
-            .map(Into::into)
-    }
-
-    /// Subtracts given [IntervalDayTime] to the current datetime.
-    pub fn sub_day_time(&self, interval: IntervalDayTime) -> Option<Self> {
-        let naive_datetime = self.to_chrono_datetime()?;
-
-        naive_datetime
-            .checked_sub_days(Days::new(interval.days as u64))?
-            .checked_sub_signed(TimeDelta::milliseconds(interval.milliseconds as i64))
-            .map(Into::into)
-    }
-
-    /// Subtracts given [IntervalMonthDayNano] to the current datetime.
-    pub fn sub_month_day_nano(&self, interval: IntervalMonthDayNano) -> Option<Self> {
-        let naive_datetime = self.to_chrono_datetime()?;
-
-        naive_datetime
-            .checked_sub_months(Months::new(interval.months as u32))?
-            .checked_sub_days(Days::new(interval.days as u64))?
-            .checked_sub_signed(TimeDelta::nanoseconds(interval.nanoseconds))
-            .map(Into::into)
-    }
-
-    /// Convert to [common_time::date].
-    pub fn to_date(&self) -> Option<Date> {
-        self.to_chrono_datetime().map(|d| Date::from(d.date()))
-    }
-
-    pub fn negative(&self) -> Self {
-        Self(-self.0)
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-    use crate::timezone::set_default_timezone;
-
-    #[test]
-    pub fn test_new_date_time() {
-        set_default_timezone(Some("Asia/Shanghai")).unwrap();
-        assert_eq!("1970-01-01 08:00:00+0800", DateTime::new(0).to_string());
-        assert_eq!("1970-01-01 08:00:01+0800", DateTime::new(1000).to_string());
-        assert_eq!("1970-01-01 07:59:59+0800", DateTime::new(-1000).to_string());
-    }
-
-    #[test]
-    pub fn test_parse_from_string() {
-        set_default_timezone(Some("Asia/Shanghai")).unwrap();
-        let time = "1970-01-01 00:00:00+0800";
-        let dt = DateTime::from_str(time, None).unwrap();
-        assert_eq!(time, &dt.to_string());
-        let dt = DateTime::from_str("      1970-01-01       00:00:00+0800       ", None).unwrap();
-        assert_eq!(time, &dt.to_string());
-    }
-
-    #[test]
-    pub fn test_from() {
-        let d: DateTime = 42.into();
-        assert_eq!(42, d.val());
-    }
-
-    #[test]
-    fn test_add_sub_interval() {
-        let datetime = DateTime::new(1000);
-
-        let interval = IntervalDayTime::new(1, 200);
-
-        let new_datetime = datetime.add_day_time(interval).unwrap();
-        assert_eq!(new_datetime.val(), 1000 + 3600 * 24 * 1000 + 200);
-
-        assert_eq!(datetime, new_datetime.sub_day_time(interval).unwrap());
-    }
-
-    #[test]
-    fn test_parse_local_date_time() {
-        set_default_timezone(Some("Asia/Shanghai")).unwrap();
-        assert_eq!(
-            -28800000,
-            DateTime::from_str("1970-01-01 00:00:00", None)
-                .unwrap()
-                .val()
-        );
-        assert_eq!(
-            0,
-            DateTime::from_str("1970-01-01 08:00:00", None)
-                .unwrap()
-                .val()
-        );
-        assert_eq!(
-            42,
-            DateTime::from_str("1970-01-01 08:00:00.042", None)
-                .unwrap()
-                .val()
-        );
-        assert_eq!(
-            42,
-            DateTime::from_str("1970-01-01 08:00:00.042424", None)
-                .unwrap()
-                .val()
-        );
-
-        assert_eq!(
-            0,
-            DateTime::from_str(
-                "1970-01-01 08:00:00",
-                Some(&Timezone::from_tz_string("Asia/Shanghai").unwrap())
-            )
-            .unwrap()
-            .val()
-        );
-
-        assert_eq!(
-            -28800000,
-            DateTime::from_str(
-                "1970-01-01 00:00:00",
-                Some(&Timezone::from_tz_string("Asia/Shanghai").unwrap())
-            )
-            .unwrap()
-            .val()
-        );
-
-        assert_eq!(
-            28800000,
-            DateTime::from_str(
-                "1970-01-01 00:00:00",
-                Some(&Timezone::from_tz_string("-8:00").unwrap())
-            )
-            .unwrap()
-            .val()
-        );
-    }
-
-    #[test]
-    fn test_parse_local_date_time_with_tz() {
-        let ts = DateTime::from_str("1970-01-01 08:00:00+0000", None)
-            .unwrap()
-            .val();
-        assert_eq!(28800000, ts);
-        let ts = DateTime::from_str("1970-01-01 00:00:00.042+0000", None)
-            .unwrap()
-            .val();
-        assert_eq!(42, ts);
-
-        // the string has the time zone info, the argument doesn't change the result
-        let ts = DateTime::from_str(
-            "1970-01-01 08:00:00+0000",
-            Some(&Timezone::from_tz_string("-8:00").unwrap()),
-        )
-        .unwrap()
-        .val();
-        assert_eq!(28800000, ts);
-    }
-
-    #[test]
-    fn test_as_formatted_string() {
-        let d: DateTime = DateTime::new(1000);
-
-        assert_eq!(
-            "1970-01-01",
-            d.as_formatted_string("%Y-%m-%d", None).unwrap().unwrap()
-        );
-        assert_eq!(
-            "1970-01-01 00:00:01",
-            d.as_formatted_string("%Y-%m-%d %H:%M:%S", None)
-                .unwrap()
-                .unwrap()
-        );
-        assert_eq!(
-            "1970-01-01T00:00:01:000",
-            d.as_formatted_string("%Y-%m-%dT%H:%M:%S:%3f", None)
-                .unwrap()
-                .unwrap()
-        );
-
-        assert_eq!(
-            "1970-01-01T08:00:01:000",
-            d.as_formatted_string(
-                "%Y-%m-%dT%H:%M:%S:%3f",
-                Some(&Timezone::from_tz_string("Asia/Shanghai").unwrap())
-            )
-            .unwrap()
-            .unwrap()
-        );
-    }
-
-    #[test]
-    fn test_from_max_date() {
-        let date = Date::new(i32::MAX);
-        let datetime = DateTime::from(date);
-        assert_eq!(datetime.val(), 185542587100800000);
-    }
-
-    #[test]
-    fn test_conversion_between_datetime_and_chrono_datetime() {
-        let cases = [1, 10, 100, 1000, 100000];
-        for case in cases {
-            let dt = DateTime::new(case);
-            let ndt = dt.to_chrono_datetime().unwrap();
-            let dt2 = DateTime::from(ndt);
-            assert_eq!(dt, dt2);
-        }
-    }
-}
--- a/src/common/time/src/lib.rs
+++ b/src/common/time/src/lib.rs
@@ -13,7 +13,6 @@
 // limitations under the License.

 pub mod date;
-pub mod datetime;
 pub mod duration;
 pub mod error;
 pub mod interval;
@@ -26,7 +25,6 @@ pub mod ttl;
 pub mod util;

 pub use date::Date;
-pub use datetime::DateTime;
 pub use duration::Duration;
 pub use interval::{IntervalDayTime, IntervalMonthDayNano, IntervalYearMonth};
 pub use range::RangeMillis;
--- a/src/datanode/src/config.rs
+++ b/src/datanode/src/config.rs
@@ -36,7 +36,7 @@ use servers::Mode;
 pub const DEFAULT_OBJECT_STORE_CACHE_SIZE: ReadableSize = ReadableSize::gb(5);

 /// Default data home in file storage
-const DEFAULT_DATA_HOME: &str = "/tmp/greptimedb";
+const DEFAULT_DATA_HOME: &str = "./greptimedb_data";

 /// Object storage config
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
@@ -171,6 +171,10 @@ pub struct S3Config {
    pub secret_access_key: SecretString,
    pub endpoint: Option<String>,
    pub region: Option<String>,
+    /// Enable virtual host style so that opendal will send API requests in virtual host style instead of path style.
+    /// By default, opendal will send API to https://s3.us-east-1.amazonaws.com/bucket_name
+    /// Enabled, opendal will send API to https://bucket_name.s3.us-east-1.amazonaws.com
+    pub enable_virtual_host_style: bool,
    #[serde(flatten)]
    pub cache: ObjectStorageCacheConfig,
    pub http_client: HttpClientConfig,
@@ -185,6 +189,7 @@ impl PartialEq for S3Config {
            && self.secret_access_key.expose_secret() == other.secret_access_key.expose_secret()
            && self.endpoint == other.endpoint
            && self.region == other.region
+            && self.enable_virtual_host_style == other.enable_virtual_host_style
            && self.cache == other.cache
            && self.http_client == other.http_client
    }
@@ -289,6 +294,7 @@ impl Default for S3Config {
            root: String::default(),
            access_key_id: SecretString::from(String::default()),
            secret_access_key: SecretString::from(String::default()),
+            enable_virtual_host_style: false,
            endpoint: Option::default(),
            region: Option::default(),
            cache: ObjectStorageCacheConfig::default(),
--- a/src/datanode/src/store/s3.rs
+++ b/src/datanode/src/store/s3.rs
@@ -41,10 +41,13 @@ pub(crate) async fn new_s3_object_store(s3_config: &S3Config) -> Result<ObjectSt

    if s3_config.endpoint.is_some() {
        builder = builder.endpoint(s3_config.endpoint.as_ref().unwrap());
-    };
+    }
    if s3_config.region.is_some() {
        builder = builder.region(s3_config.region.as_ref().unwrap());
-    };
+    }
+    if s3_config.enable_virtual_host_style {
+        builder = builder.enable_virtual_host_style();
+    }

    Ok(ObjectStore::new(builder)
        .context(error::InitBackendSnafu)?
--- a/src/datatypes/src/data_type.rs
+++ b/src/datatypes/src/data_type.rs
@@ -30,13 +30,13 @@ use serde::{Deserialize, Serialize};
 use crate::error::{self, Error, Result};
 use crate::type_id::LogicalTypeId;
 use crate::types::{
-    BinaryType, BooleanType, DateTimeType, DateType, Decimal128Type, DictionaryType,
-    DurationMicrosecondType, DurationMillisecondType, DurationNanosecondType, DurationSecondType,
-    DurationType, Float32Type, Float64Type, Int16Type, Int32Type, Int64Type, Int8Type,
-    IntervalDayTimeType, IntervalMonthDayNanoType, IntervalType, IntervalYearMonthType, JsonType,
-    ListType, NullType, StringType, TimeMillisecondType, TimeType, TimestampMicrosecondType,
-    TimestampMillisecondType, TimestampNanosecondType, TimestampSecondType, TimestampType,
-    UInt16Type, UInt32Type, UInt64Type, UInt8Type, VectorType,
+    BinaryType, BooleanType, DateType, Decimal128Type, DictionaryType, DurationMicrosecondType,
+    DurationMillisecondType, DurationNanosecondType, DurationSecondType, DurationType, Float32Type,
+    Float64Type, Int16Type, Int32Type, Int64Type, Int8Type, IntervalDayTimeType,
+    IntervalMonthDayNanoType, IntervalType, IntervalYearMonthType, JsonType, ListType, NullType,
+    StringType, TimeMillisecondType, TimeType, TimestampMicrosecondType, TimestampMillisecondType,
+    TimestampNanosecondType, TimestampSecondType, TimestampType, UInt16Type, UInt32Type,
+    UInt64Type, UInt8Type, VectorType,
 };
 use crate::value::Value;
 use crate::vectors::MutableVector;
@@ -68,7 +68,6 @@ pub enum ConcreteDataType {

    // Date and time types:
    Date(DateType),
-    DateTime(DateTimeType),
    Timestamp(TimestampType),
    Time(TimeType),

@@ -107,7 +106,6 @@ impl fmt::Display for ConcreteDataType {
            ConcreteDataType::Binary(v) => write!(f, "{}", v.name()),
            ConcreteDataType::String(v) => write!(f, "{}", v.name()),
            ConcreteDataType::Date(v) => write!(f, "{}", v.name()),
-            ConcreteDataType::DateTime(v) => write!(f, "{}", v.name()),
            ConcreteDataType::Timestamp(t) => match t {
                TimestampType::Second(v) => write!(f, "{}", v.name()),
                TimestampType::Millisecond(v) => write!(f, "{}", v.name()),
@@ -163,7 +161,6 @@ impl ConcreteDataType {
            self,
            ConcreteDataType::String(_)
                | ConcreteDataType::Date(_)
-                | ConcreteDataType::DateTime(_)
                | ConcreteDataType::Timestamp(_)
                | ConcreteDataType::Time(_)
                | ConcreteDataType::Interval(_)
@@ -183,7 +180,6 @@ impl ConcreteDataType {
                | ConcreteDataType::Int32(_)
                | ConcreteDataType::Int64(_)
                | ConcreteDataType::Date(_)
-                | ConcreteDataType::DateTime(_)
                | ConcreteDataType::Timestamp(_)
                | ConcreteDataType::Time(_)
                | ConcreteDataType::Interval(_)
@@ -385,7 +381,7 @@ impl ConcreteDataType {
            &ConcreteDataType::Binary(_) | &ConcreteDataType::Vector(_) => "BYTEA",
            &ConcreteDataType::String(_) => "VARCHAR",
            &ConcreteDataType::Date(_) => "DATE",
-            &ConcreteDataType::DateTime(_) | &ConcreteDataType::Timestamp(_) => "TIMESTAMP",
+            &ConcreteDataType::Timestamp(_) => "TIMESTAMP",
            &ConcreteDataType::Time(_) => "TIME",
            &ConcreteDataType::Interval(_) => "INTERVAL",
            &ConcreteDataType::Decimal128(_) => "NUMERIC",
@@ -402,7 +398,7 @@ impl ConcreteDataType {
                &ConcreteDataType::Binary(_) => "_BYTEA",
                &ConcreteDataType::String(_) => "_VARCHAR",
                &ConcreteDataType::Date(_) => "_DATE",
-                &ConcreteDataType::DateTime(_) | &ConcreteDataType::Timestamp(_) => "_TIMESTAMP",
+                &ConcreteDataType::Timestamp(_) => "_TIMESTAMP",
                &ConcreteDataType::Time(_) => "_TIME",
                &ConcreteDataType::Interval(_) => "_INTERVAL",
                &ConcreteDataType::Decimal128(_) => "_NUMERIC",
@@ -441,7 +437,6 @@ impl TryFrom<&ArrowDataType> for ConcreteDataType {
            ArrowDataType::Float32 => Self::float32_datatype(),
            ArrowDataType::Float64 => Self::float64_datatype(),
            ArrowDataType::Date32 => Self::date_datatype(),
-            ArrowDataType::Date64 => Self::datetime_datatype(),
            ArrowDataType::Timestamp(u, _) => ConcreteDataType::from_arrow_time_unit(u),
            ArrowDataType::Interval(u) => ConcreteDataType::from_arrow_interval_unit(u),
            ArrowDataType::Binary | ArrowDataType::LargeBinary => Self::binary_datatype(),
@@ -490,7 +485,7 @@ macro_rules! impl_new_concrete_type_functions {

 impl_new_concrete_type_functions!(
    Null, Boolean, UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64,
-    Binary, Date, DateTime, String, Json
+    Binary, Date, String, Json
 );

 impl ConcreteDataType {
@@ -814,7 +809,6 @@ mod tests {
        assert!(ConcreteDataType::string_datatype().is_stringifiable());
        assert!(ConcreteDataType::binary_datatype().is_stringifiable());
        assert!(ConcreteDataType::date_datatype().is_stringifiable());
-        assert!(ConcreteDataType::datetime_datatype().is_stringifiable());
        assert!(ConcreteDataType::timestamp_second_datatype().is_stringifiable());
        assert!(ConcreteDataType::timestamp_millisecond_datatype().is_stringifiable());
        assert!(ConcreteDataType::timestamp_microsecond_datatype().is_stringifiable());
@@ -843,7 +837,6 @@ mod tests {
        assert!(ConcreteDataType::int32_datatype().is_signed());
        assert!(ConcreteDataType::int64_datatype().is_signed());
        assert!(ConcreteDataType::date_datatype().is_signed());
-        assert!(ConcreteDataType::datetime_datatype().is_signed());
        assert!(ConcreteDataType::timestamp_second_datatype().is_signed());
        assert!(ConcreteDataType::timestamp_millisecond_datatype().is_signed());
        assert!(ConcreteDataType::timestamp_microsecond_datatype().is_signed());
@@ -878,7 +871,6 @@ mod tests {
        assert!(!ConcreteDataType::int32_datatype().is_unsigned());
        assert!(!ConcreteDataType::int64_datatype().is_unsigned());
        assert!(!ConcreteDataType::date_datatype().is_unsigned());
-        assert!(!ConcreteDataType::datetime_datatype().is_unsigned());
        assert!(!ConcreteDataType::timestamp_second_datatype().is_unsigned());
        assert!(!ConcreteDataType::timestamp_millisecond_datatype().is_unsigned());
        assert!(!ConcreteDataType::timestamp_microsecond_datatype().is_unsigned());
--- a/src/datatypes/src/scalars.rs
+++ b/src/datatypes/src/scalars.rs
@@ -15,7 +15,7 @@
 use std::any::Any;

 use common_decimal::Decimal128;
-use common_time::{Date, DateTime};
+use common_time::Date;

 use crate::types::{
    Float32Type, Float64Type, Int16Type, Int32Type, Int64Type, Int8Type, UInt16Type, UInt32Type,
@@ -23,8 +23,8 @@ use crate::types::{
 };
 use crate::value::{ListValue, ListValueRef, Value};
 use crate::vectors::{
-    BinaryVector, BooleanVector, DateTimeVector, DateVector, Decimal128Vector, ListVector,
-    MutableVector, PrimitiveVector, StringVector, Vector,
+    BinaryVector, BooleanVector, DateVector, Decimal128Vector, ListVector, MutableVector,
+    PrimitiveVector, StringVector, Vector,
 };

 fn get_iter_capacity<T, I: Iterator<Item = T>>(iter: &I) -> usize {
@@ -302,27 +302,6 @@ impl ScalarRef<'_> for Decimal128 {
    }
 }

-impl Scalar for DateTime {
-    type VectorType = DateTimeVector;
-    type RefType<'a> = DateTime;
-
-    fn as_scalar_ref(&self) -> Self::RefType<'_> {
-        *self
-    }
-
-    fn upcast_gat<'short, 'long: 'short>(long: Self::RefType<'long>) -> Self::RefType<'short> {
-        long
-    }
-}
-
-impl ScalarRef<'_> for DateTime {
-    type ScalarType = DateTime;
-
-    fn to_owned_scalar(&self) -> Self::ScalarType {
-        *self
-    }
-}
-
 // Timestamp types implement Scalar and ScalarRef in `src/timestamp.rs`.

 impl Scalar for ListValue {
@@ -428,13 +407,6 @@ mod tests {
        assert_eq!(decimal, decimal.to_owned_scalar());
    }

-    #[test]
-    fn test_datetime_scalar() {
-        let dt = DateTime::new(123);
-        assert_eq!(dt, dt.as_scalar_ref());
-        assert_eq!(dt, dt.to_owned_scalar());
-    }
-
    #[test]
    fn test_list_value_scalar() {
        let list_value =
--- a/src/datatypes/src/type_id.rs
+++ b/src/datatypes/src/type_id.rs
@@ -40,9 +40,6 @@ pub enum LogicalTypeId {
    /// Date representing the elapsed time since UNIX epoch (1970-01-01)
    /// in days (32 bits).
    Date,
-    /// Datetime representing the elapsed time since UNIX epoch (1970-01-01) in
-    /// seconds/milliseconds/microseconds/nanoseconds, determined by precision.
-    DateTime,

    TimestampSecond,
    TimestampMillisecond,
@@ -100,7 +97,6 @@ impl LogicalTypeId {
            LogicalTypeId::String => ConcreteDataType::string_datatype(),
            LogicalTypeId::Binary => ConcreteDataType::binary_datatype(),
            LogicalTypeId::Date => ConcreteDataType::date_datatype(),
-            LogicalTypeId::DateTime => ConcreteDataType::datetime_datatype(),
            LogicalTypeId::TimestampSecond => ConcreteDataType::timestamp_second_datatype(),
            LogicalTypeId::TimestampMillisecond => {
                ConcreteDataType::timestamp_millisecond_datatype()
--- a/src/datatypes/src/types.rs
+++ b/src/datatypes/src/types.rs
@@ -16,7 +16,6 @@ mod binary_type;
 mod boolean_type;
 pub mod cast;
 mod date_type;
-mod datetime_type;
 mod decimal_type;
 mod dictionary_type;
 mod duration_type;
@@ -34,7 +33,6 @@ pub use binary_type::BinaryType;
 pub use boolean_type::BooleanType;
 pub use cast::{cast, cast_with_opt};
 pub use date_type::DateType;
-pub use datetime_type::DateTimeType;
 pub use decimal_type::Decimal128Type;
 pub use dictionary_type::DictionaryType;
 pub use duration_type::{
--- a/src/datatypes/src/types/cast.rs
+++ b/src/datatypes/src/types/cast.rs
@@ -119,10 +119,6 @@ pub fn can_cast_type(src_value: &Value, dest_type: &ConcreteDataType) -> bool {
        (Date(_), Int32(_) | Timestamp(_) | String(_)) => true,
        (Int32(_) | String(_) | Timestamp(_), Date(_)) => true,
        (Date(_), Date(_)) => true,
-        // DateTime type
-        (DateTime(_), Int64(_) | Timestamp(_) | String(_)) => true,
-        (Int64(_) | Timestamp(_) | String(_), DateTime(_)) => true,
-        (DateTime(_), DateTime(_)) => true,
        // Timestamp type
        (Timestamp(_), Int64(_) | String(_)) => true,
        (Int64(_) | String(_), Timestamp(_)) => true,
@@ -175,7 +171,7 @@ mod tests {
    use common_base::bytes::StringBytes;
    use common_time::time::Time;
    use common_time::timezone::set_default_timezone;
-    use common_time::{Date, DateTime, Timestamp};
+    use common_time::{Date, Timestamp};
    use ordered_float::OrderedFloat;

    use super::*;
@@ -274,7 +270,6 @@ mod tests {
            null_datatype,
            boolean_datatype,
            date_datatype,
-            datetime_datatype,
            timestamp_second_datatype,
            binary_datatype
        );
@@ -287,23 +282,12 @@ mod tests {
            timestamp_second_datatype,
            string_datatype
        );
-
-        // datetime -> other types
-        test_can_cast!(
-            Value::DateTime(DateTime::from_str_system("2021-01-01 00:00:00").unwrap()),
-            null_datatype,
-            int64_datatype,
-            timestamp_second_datatype,
-            string_datatype
-        );
-
        // timestamp -> other types
        test_can_cast!(
            Value::Timestamp(Timestamp::from_str_utc("2021-01-01 00:00:00").unwrap()),
            null_datatype,
            int64_datatype,
            date_datatype,
-            datetime_datatype,
            string_datatype
        );

--- a/src/datatypes/src/types/date_type.rs
+++ b/src/datatypes/src/types/date_type.rs
@@ -55,7 +55,6 @@ impl DataType for DateType {
            Value::Int32(v) => Some(Value::Date(Date::from(v))),
            Value::String(v) => Date::from_str_utc(v.as_utf8()).map(Value::Date).ok(),
            Value::Timestamp(v) => v.to_chrono_date().map(|date| Value::Date(date.into())),
-            Value::DateTime(v) => Some(Value::DateTime(v)),
            _ => None,
        }
    }
--- a/src/datatypes/src/types/datetime_type.rs
+++ b/src/datatypes/src/types/datetime_type.rs
@@ -1,140 +0,0 @@
-// 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 arrow::datatypes::{DataType as ArrowDataType, Date64Type};
-use common_time::DateTime;
-use serde::{Deserialize, Serialize};
-use snafu::OptionExt;
-
-use crate::data_type::{ConcreteDataType, DataType};
-use crate::error::{self, Result};
-use crate::prelude::{LogicalTypeId, MutableVector, ScalarVectorBuilder, Value, ValueRef, Vector};
-use crate::types::LogicalPrimitiveType;
-use crate::vectors::{DateTimeVector, DateTimeVectorBuilder, PrimitiveVector};
-
-const MILLISECOND_VARIATION: u64 = 3;
-/// Data type for [`DateTime`].
-#[derive(Debug, Default, Clone, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
-pub struct DateTimeType;
-
-impl DateTimeType {
-    pub fn precision(&self) -> u64 {
-        MILLISECOND_VARIATION
-    }
-}
-
-impl DataType for DateTimeType {
-    fn name(&self) -> String {
-        "DateTime".to_string()
-    }
-
-    fn logical_type_id(&self) -> LogicalTypeId {
-        LogicalTypeId::DateTime
-    }
-
-    fn default_value(&self) -> Value {
-        Value::DateTime(DateTime::default())
-    }
-
-    fn as_arrow_type(&self) -> ArrowDataType {
-        ArrowDataType::Date64
-    }
-
-    fn create_mutable_vector(&self, capacity: usize) -> Box<dyn MutableVector> {
-        Box::new(DateTimeVectorBuilder::with_capacity(capacity))
-    }
-
-    fn try_cast(&self, from: Value) -> Option<Value> {
-        match from {
-            Value::Int64(v) => Some(Value::DateTime(DateTime::from(v))),
-            Value::Timestamp(v) => v.to_chrono_datetime().map(|d| Value::DateTime(d.into())),
-            Value::String(v) => DateTime::from_str_system(v.as_utf8())
-                .map(Value::DateTime)
-                .ok(),
-            _ => None,
-        }
-    }
-}
-
-impl LogicalPrimitiveType for DateTimeType {
-    type ArrowPrimitive = Date64Type;
-    type Native = i64;
-    type Wrapper = DateTime;
-    type LargestType = Self;
-
-    fn build_data_type() -> ConcreteDataType {
-        ConcreteDataType::datetime_datatype()
-    }
-
-    fn type_name() -> &'static str {
-        "DateTime"
-    }
-
-    fn cast_vector(vector: &dyn Vector) -> Result<&PrimitiveVector<Self>> {
-        vector
-            .as_any()
-            .downcast_ref::<DateTimeVector>()
-            .with_context(|| error::CastTypeSnafu {
-                msg: format!(
-                    "Failed to cast {} to DateTimeVector",
-                    vector.vector_type_name()
-                ),
-            })
-    }
-
-    fn cast_value_ref(value: ValueRef) -> Result<Option<Self::Wrapper>> {
-        match value {
-            ValueRef::Null => Ok(None),
-            ValueRef::DateTime(v) => Ok(Some(v)),
-            other => error::CastTypeSnafu {
-                msg: format!("Failed to cast value {other:?} to DateTime"),
-            }
-            .fail(),
-        }
-    }
-}
-
-#[cfg(test)]
-mod tests {
-
-    use common_time::timezone::set_default_timezone;
-    use common_time::Timestamp;
-
-    use super::*;
-
-    #[test]
-    fn test_datetime_cast() {
-        // cast from Int64
-        let val = Value::Int64(1000);
-        let dt = ConcreteDataType::datetime_datatype().try_cast(val).unwrap();
-        assert_eq!(dt, Value::DateTime(DateTime::from(1000)));
-
-        // cast from String
-        set_default_timezone(Some("Asia/Shanghai")).unwrap();
-        let val = Value::String("1970-01-01 00:00:00+0800".into());
-        let dt = ConcreteDataType::datetime_datatype().try_cast(val).unwrap();
-        assert_eq!(
-            dt,
-            Value::DateTime(DateTime::from_str_system("1970-01-01 00:00:00+0800").unwrap())
-        );
-
-        // cast from Timestamp
-        let val = Value::Timestamp(Timestamp::from_str_utc("2020-09-08 21:42:29+0800").unwrap());
-        let dt = ConcreteDataType::datetime_datatype().try_cast(val).unwrap();
-        assert_eq!(
-            dt,
-            Value::DateTime(DateTime::from_str_system("2020-09-08 21:42:29+0800").unwrap())
-        );
-    }
-}
--- a/src/datatypes/src/types/primitive_type.rs
+++ b/src/datatypes/src/types/primitive_type.rs
@@ -16,7 +16,7 @@ use std::cmp::Ordering;
 use std::fmt;

 use arrow::datatypes::{ArrowNativeType, ArrowPrimitiveType, DataType as ArrowDataType};
-use common_time::{Date, DateTime};
+use common_time::Date;
 use serde::{Deserialize, Serialize};
 use snafu::OptionExt;

@@ -25,7 +25,7 @@ use crate::error::{self, Result};
 use crate::scalars::{Scalar, ScalarRef, ScalarVectorBuilder};
 use crate::type_id::LogicalTypeId;
 use crate::types::boolean_type::bool_to_numeric;
-use crate::types::{DateTimeType, DateType};
+use crate::types::DateType;
 use crate::value::{Value, ValueRef};
 use crate::vectors::{MutableVector, PrimitiveVector, PrimitiveVectorBuilder, Vector};

@@ -157,19 +157,6 @@ impl WrapperType for Date {
    }
 }

-impl WrapperType for DateTime {
-    type LogicalType = DateTimeType;
-    type Native = i64;
-
-    fn from_native(value: Self::Native) -> Self {
-        DateTime::new(value)
-    }
-
-    fn into_native(self) -> Self::Native {
-        self.val()
-    }
-}
-
 macro_rules! define_logical_primitive_type {
    ($Native: ident, $TypeId: ident, $DataType: ident, $Largest: ident) => {
        // We need to define it as an empty struct `struct DataType {}` instead of a struct-unit
@@ -362,7 +349,6 @@ impl DataType for Int64Type {
            Value::Float32(v) => num::cast::cast(v).map(Value::Int64),
            Value::Float64(v) => num::cast::cast(v).map(Value::Int64),
            Value::String(v) => v.as_utf8().parse::<i64>().map(Value::Int64).ok(),
-            Value::DateTime(v) => Some(Value::Int64(v.val())),
            Value::Timestamp(v) => Some(Value::Int64(v.value())),
            Value::Time(v) => Some(Value::Int64(v.value())),
            // We don't allow casting interval type to int.
--- a/src/datatypes/src/types/string_type.rs
+++ b/src/datatypes/src/types/string_type.rs
@@ -75,7 +75,6 @@ impl DataType for StringType {
            Value::Float64(v) => Some(Value::String(StringBytes::from(v.to_string()))),
            Value::String(v) => Some(Value::String(v)),
            Value::Date(v) => Some(Value::String(StringBytes::from(v.to_string()))),
-            Value::DateTime(v) => Some(Value::String(StringBytes::from(v.to_string()))),
            Value::Timestamp(v) => Some(Value::String(StringBytes::from(v.to_iso8601_string()))),
            Value::Time(v) => Some(Value::String(StringBytes::from(v.to_iso8601_string()))),
            Value::IntervalYearMonth(v) => {
--- a/src/datatypes/src/types/timestamp_type.rs
+++ b/src/datatypes/src/types/timestamp_type.rs
@@ -132,7 +132,6 @@ macro_rules! impl_data_type_for_timestamp {
                        Value::Timestamp(v) => v.convert_to(TimeUnit::$unit).map(Value::Timestamp),
                        Value::String(v) => Timestamp::from_str_utc(v.as_utf8()).map(Value::Timestamp).ok(),
                        Value::Int64(v) => Some(Value::Timestamp(Timestamp::new(v, TimeUnit::$unit))),
-                        Value::DateTime(v) => Timestamp::new_second(v.val()).convert_to(TimeUnit::$unit).map(Value::Timestamp),
                        Value::Date(v) => Timestamp::new_second(v.to_secs()).convert_to(TimeUnit::$unit).map(Value::Timestamp),
                        _ => None
                    }
@@ -202,7 +201,7 @@ impl_data_type_for_timestamp!(Microsecond);
 #[cfg(test)]
 mod tests {
    use common_time::timezone::set_default_timezone;
-    use common_time::{Date, DateTime};
+    use common_time::Date;

    use super::*;

@@ -249,13 +248,6 @@ mod tests {
            .unwrap();
        assert_eq!(ts, Value::Timestamp(Timestamp::new_second(1694589525)));

-        // Datetime -> TimestampSecond
-        let dt = Value::DateTime(DateTime::from(1234567));
-        let ts = ConcreteDataType::timestamp_second_datatype()
-            .try_cast(dt)
-            .unwrap();
-        assert_eq!(ts, Value::Timestamp(Timestamp::new_second(1234567)));
-
        // Date -> TimestampMillisecond
        let d = Value::Date(Date::from_str_utc("1970-01-01").unwrap());
        let ts = ConcreteDataType::timestamp_millisecond_datatype()
--- a/src/datatypes/src/value.rs
+++ b/src/datatypes/src/value.rs
@@ -24,7 +24,6 @@ use common_base::bytes::{Bytes, StringBytes};
 use common_decimal::Decimal128;
 use common_telemetry::error;
 use common_time::date::Date;
-use common_time::datetime::DateTime;
 use common_time::interval::IntervalUnit;
 use common_time::time::Time;
 use common_time::timestamp::{TimeUnit, Timestamp};
@@ -75,7 +74,6 @@ pub enum Value {

    // Date & Time types:
    Date(Date),
-    DateTime(DateTime),
    Timestamp(Timestamp),
    Time(Time),
    Duration(Duration),
@@ -112,7 +110,6 @@ impl Display for Value {
                write!(f, "{hex}")
            }
            Value::Date(v) => write!(f, "{v}"),
-            Value::DateTime(v) => write!(f, "{v}"),
            Value::Timestamp(v) => write!(f, "{}", v.to_iso8601_string()),
            Value::Time(t) => write!(f, "{}", t.to_iso8601_string()),
            Value::IntervalYearMonth(v) => {
@@ -162,7 +159,6 @@ macro_rules! define_data_type_func {
                $struct::String(_) => ConcreteDataType::string_datatype(),
                $struct::Binary(_) => ConcreteDataType::binary_datatype(),
                $struct::Date(_) => ConcreteDataType::date_datatype(),
-                $struct::DateTime(_) => ConcreteDataType::datetime_datatype(),
                $struct::Time(t) => ConcreteDataType::time_datatype(*t.unit()),
                $struct::Timestamp(v) => ConcreteDataType::timestamp_datatype(v.unit()),
                $struct::IntervalYearMonth(_) => {
@@ -222,7 +218,6 @@ impl Value {
            Value::String(v) => ValueRef::String(v.as_utf8()),
            Value::Binary(v) => ValueRef::Binary(v),
            Value::Date(v) => ValueRef::Date(*v),
-            Value::DateTime(v) => ValueRef::DateTime(*v),
            Value::List(v) => ValueRef::List(ListValueRef::Ref { val: v }),
            Value::Timestamp(v) => ValueRef::Timestamp(*v),
            Value::Time(v) => ValueRef::Time(*v),
@@ -258,14 +253,6 @@ impl Value {
        }
    }

-    /// Cast Value to DateTime. Return None if value is not a valid datetime data type.
-    pub fn as_datetime(&self) -> Option<DateTime> {
-        match self {
-            Value::DateTime(t) => Some(*t),
-            _ => None,
-        }
-    }
-
    /// Cast Value to [Time]. Return None if value is not a valid time data type.
    pub fn as_time(&self) -> Option<Time> {
        match self {
@@ -298,6 +285,20 @@ impl Value {
        }
    }

+    /// Cast Value to i64. Return None if value is not a valid int64 data type.
+    pub fn as_i64(&self) -> Option<i64> {
+        match self {
+            Value::Int8(v) => Some(*v as _),
+            Value::Int16(v) => Some(*v as _),
+            Value::Int32(v) => Some(*v as _),
+            Value::Int64(v) => Some(*v),
+            Value::UInt8(v) => Some(*v as _),
+            Value::UInt16(v) => Some(*v as _),
+            Value::UInt32(v) => Some(*v as _),
+            _ => None,
+        }
+    }
+
    /// Cast Value to u64. Return None if value is not a valid uint64 data type.
    pub fn as_u64(&self) -> Option<u64> {
        match self {
@@ -308,7 +309,6 @@ impl Value {
            _ => None,
        }
    }
-
    /// Cast Value to f64. Return None if it's not castable;
    pub fn as_f64_lossy(&self) -> Option<f64> {
        match self {
@@ -345,7 +345,6 @@ impl Value {
            Value::Binary(_) => LogicalTypeId::Binary,
            Value::List(_) => LogicalTypeId::List,
            Value::Date(_) => LogicalTypeId::Date,
-            Value::DateTime(_) => LogicalTypeId::DateTime,
            Value::Timestamp(t) => match t.unit() {
                TimeUnit::Second => LogicalTypeId::TimestampSecond,
                TimeUnit::Millisecond => LogicalTypeId::TimestampMillisecond,
@@ -401,7 +400,6 @@ impl Value {
            Value::String(v) => ScalarValue::Utf8(Some(v.as_utf8().to_string())),
            Value::Binary(v) => ScalarValue::Binary(Some(v.to_vec())),
            Value::Date(v) => ScalarValue::Date32(Some(v.val())),
-            Value::DateTime(v) => ScalarValue::Date64(Some(v.val())),
            Value::Null => to_null_scalar_value(output_type)?,
            Value::List(list) => {
                // Safety: The logical type of the value and output_type are the same.
@@ -463,7 +461,6 @@ impl Value {
            Value::Float64(x) => Some(Value::Float64(-*x)),
            Value::Decimal128(x) => Some(Value::Decimal128(x.negative())),
            Value::Date(x) => Some(Value::Date(x.negative())),
-            Value::DateTime(x) => Some(Value::DateTime(x.negative())),
            Value::Timestamp(x) => Some(Value::Timestamp(x.negative())),
            Value::Time(x) => Some(Value::Time(x.negative())),
            Value::Duration(x) => Some(Value::Duration(x.negative())),
@@ -525,7 +522,6 @@ pub fn to_null_scalar_value(output_type: &ConcreteDataType) -> Result<ScalarValu
        }
        ConcreteDataType::String(_) => ScalarValue::Utf8(None),
        ConcreteDataType::Date(_) => ScalarValue::Date32(None),
-        ConcreteDataType::DateTime(_) => ScalarValue::Date64(None),
        ConcreteDataType::Timestamp(t) => timestamp_to_scalar_value(t.unit(), None),
        ConcreteDataType::Interval(v) => match v {
            IntervalType::YearMonth(_) => ScalarValue::IntervalYearMonth(None),
@@ -631,7 +627,6 @@ macro_rules! impl_ord_for_value_like {
                ($Type::String(v1), $Type::String(v2)) => v1.cmp(v2),
                ($Type::Binary(v1), $Type::Binary(v2)) => v1.cmp(v2),
                ($Type::Date(v1), $Type::Date(v2)) => v1.cmp(v2),
-                ($Type::DateTime(v1), $Type::DateTime(v2)) => v1.cmp(v2),
                ($Type::Timestamp(v1), $Type::Timestamp(v2)) => v1.cmp(v2),
                ($Type::Time(v1), $Type::Time(v2)) => v1.cmp(v2),
                ($Type::IntervalYearMonth(v1), $Type::IntervalYearMonth(v2)) => v1.cmp(v2),
@@ -712,7 +707,6 @@ impl_try_from_value!(String, StringBytes);
 impl_try_from_value!(Binary, Bytes);
 impl_try_from_value!(Date, Date);
 impl_try_from_value!(Time, Time);
-impl_try_from_value!(DateTime, DateTime);
 impl_try_from_value!(Timestamp, Timestamp);
 impl_try_from_value!(IntervalYearMonth, IntervalYearMonth);
 impl_try_from_value!(IntervalDayTime, IntervalDayTime);
@@ -756,7 +750,6 @@ impl_value_from!(String, StringBytes);
 impl_value_from!(Binary, Bytes);
 impl_value_from!(Date, Date);
 impl_value_from!(Time, Time);
-impl_value_from!(DateTime, DateTime);
 impl_value_from!(Timestamp, Timestamp);
 impl_value_from!(IntervalYearMonth, IntervalYearMonth);
 impl_value_from!(IntervalDayTime, IntervalDayTime);
@@ -803,7 +796,6 @@ impl TryFrom<Value> for serde_json::Value {
            Value::String(bytes) => serde_json::Value::String(bytes.into_string()),
            Value::Binary(bytes) => serde_json::to_value(bytes)?,
            Value::Date(v) => serde_json::Value::Number(v.val().into()),
-            Value::DateTime(v) => serde_json::Value::Number(v.val().into()),
            Value::List(v) => serde_json::to_value(v)?,
            Value::Timestamp(v) => serde_json::to_value(v.value())?,
            Value::Time(v) => serde_json::to_value(v.value())?,
@@ -933,9 +925,6 @@ impl TryFrom<ScalarValue> for Value {
                Value::List(ListValue::new(items, datatype))
            }
            ScalarValue::Date32(d) => d.map(|x| Value::Date(Date::new(x))).unwrap_or(Value::Null),
-            ScalarValue::Date64(d) => d
-                .map(|x| Value::DateTime(DateTime::new(x)))
-                .unwrap_or(Value::Null),
            ScalarValue::TimestampSecond(t, _) => t
                .map(|x| Value::Timestamp(Timestamp::new(x, TimeUnit::Second)))
                .unwrap_or(Value::Null),
@@ -994,7 +983,8 @@ impl TryFrom<ScalarValue> for Value {
            | ScalarValue::Float16(_)
            | ScalarValue::Utf8View(_)
            | ScalarValue::BinaryView(_)
-            | ScalarValue::Map(_) => {
+            | ScalarValue::Map(_)
+            | ScalarValue::Date64(_) => {
                return error::UnsupportedArrowTypeSnafu {
                    arrow_type: v.data_type(),
                }
@@ -1023,7 +1013,6 @@ impl From<ValueRef<'_>> for Value {
            ValueRef::String(v) => Value::String(v.into()),
            ValueRef::Binary(v) => Value::Binary(v.into()),
            ValueRef::Date(v) => Value::Date(v),
-            ValueRef::DateTime(v) => Value::DateTime(v),
            ValueRef::Timestamp(v) => Value::Timestamp(v),
            ValueRef::Time(v) => Value::Time(v),
            ValueRef::IntervalYearMonth(v) => Value::IntervalYearMonth(v),
@@ -1063,7 +1052,6 @@ pub enum ValueRef<'a> {

    // Date & Time types:
    Date(Date),
-    DateTime(DateTime),
    Timestamp(Timestamp),
    Time(Time),
    Duration(Duration),
@@ -1175,11 +1163,6 @@ impl<'a> ValueRef<'a> {
        impl_as_for_value_ref!(self, Date)
    }

-    /// Cast itself to [DateTime].
-    pub fn as_datetime(&self) -> Result<Option<DateTime>> {
-        impl_as_for_value_ref!(self, DateTime)
-    }
-
    /// Cast itself to [Timestamp].
    pub fn as_timestamp(&self) -> Result<Option<Timestamp>> {
        impl_as_for_value_ref!(self, Timestamp)
@@ -1263,7 +1246,6 @@ impl_value_ref_from!(Int64, i64);
 impl_value_ref_from!(Float32, f32);
 impl_value_ref_from!(Float64, f64);
 impl_value_ref_from!(Date, Date);
-impl_value_ref_from!(DateTime, DateTime);
 impl_value_ref_from!(Timestamp, Timestamp);
 impl_value_ref_from!(Time, Time);
 impl_value_ref_from!(IntervalYearMonth, IntervalYearMonth);
@@ -1327,7 +1309,6 @@ pub fn transform_value_ref_to_json_value<'a>(
            }
        }
        ValueRef::Date(v) => serde_json::Value::Number(v.val().into()),
-        ValueRef::DateTime(v) => serde_json::Value::Number(v.val().into()),
        ValueRef::List(v) => serde_json::to_value(v)?,
        ValueRef::Timestamp(v) => serde_json::to_value(v.value())?,
        ValueRef::Time(v) => serde_json::to_value(v.value())?,
@@ -1426,7 +1407,6 @@ impl ValueRef<'_> {
            ValueRef::String(v) => std::mem::size_of_val(v),
            ValueRef::Binary(v) => std::mem::size_of_val(v),
            ValueRef::Date(_) => 4,
-            ValueRef::DateTime(_) => 8,
            ValueRef::Timestamp(_) => 16,
            ValueRef::Time(_) => 16,
            ValueRef::Duration(_) => 16,
@@ -1462,7 +1442,9 @@ pub fn column_data_to_json(data: ValueData) -> JsonValue {
            .unwrap_or(JsonValue::Null),
        ValueData::StringValue(s) => JsonValue::String(s),
        ValueData::DateValue(d) => JsonValue::String(Date::from(d).to_string()),
-        ValueData::DatetimeValue(d) => JsonValue::String(DateTime::from(d).to_string()),
+        ValueData::DatetimeValue(d) => {
+            JsonValue::String(Timestamp::new_microsecond(d).to_iso8601_string())
+        }
        ValueData::TimeSecondValue(d) => JsonValue::String(Time::new_second(d).to_iso8601_string()),
        ValueData::TimeMillisecondValue(d) => {
            JsonValue::String(Time::new_millisecond(d).to_iso8601_string())
@@ -1511,6 +1493,7 @@ mod tests {

    #[test]
    fn test_column_data_to_json() {
+        set_default_timezone(Some("Asia/Shanghai")).unwrap();
        assert_eq!(
            column_data_to_json(ValueData::BinaryValue(b"hello".to_vec())),
            JsonValue::String("aGVsbG8=".to_string())
@@ -1569,31 +1552,31 @@ mod tests {
        );
        assert_eq!(
            column_data_to_json(ValueData::DatetimeValue(456)),
-            JsonValue::String("1970-01-01 00:00:00.456+0000".to_string())
+            JsonValue::String("1970-01-01 08:00:00.000456+0800".to_string())
        );
        assert_eq!(
            column_data_to_json(ValueData::TimeSecondValue(789)),
-            JsonValue::String("00:13:09+0000".to_string())
+            JsonValue::String("08:13:09+0800".to_string())
        );
        assert_eq!(
            column_data_to_json(ValueData::TimeMillisecondValue(789)),
-            JsonValue::String("00:00:00.789+0000".to_string())
+            JsonValue::String("08:00:00.789+0800".to_string())
        );
        assert_eq!(
            column_data_to_json(ValueData::TimeMicrosecondValue(789)),
-            JsonValue::String("00:00:00.000789+0000".to_string())
+            JsonValue::String("08:00:00.000789+0800".to_string())
        );
        assert_eq!(
            column_data_to_json(ValueData::TimestampMillisecondValue(1234567890)),
-            JsonValue::String("1970-01-15 06:56:07.890+0000".to_string())
+            JsonValue::String("1970-01-15 14:56:07.890+0800".to_string())
        );
        assert_eq!(
            column_data_to_json(ValueData::TimestampNanosecondValue(1234567890123456789)),
-            JsonValue::String("2009-02-13 23:31:30.123456789+0000".to_string())
+            JsonValue::String("2009-02-14 07:31:30.123456789+0800".to_string())
        );
        assert_eq!(
            column_data_to_json(ValueData::TimestampSecondValue(1234567890)),
-            JsonValue::String("2009-02-13 23:31:30+0000".to_string())
+            JsonValue::String("2009-02-14 07:31:30+0800".to_string())
        );
        assert_eq!(
            column_data_to_json(ValueData::IntervalYearMonthValue(12)),
@@ -1758,12 +1741,6 @@ mod tests {
        );
        assert_eq!(Value::Null, ScalarValue::Date32(None).try_into().unwrap());

-        assert_eq!(
-            Value::DateTime(DateTime::new(456)),
-            ScalarValue::Date64(Some(456)).try_into().unwrap()
-        );
-        assert_eq!(Value::Null, ScalarValue::Date64(None).try_into().unwrap());
-
        assert_eq!(
            Value::Timestamp(Timestamp::new(1, TimeUnit::Second)),
            ScalarValue::TimestampSecond(Some(1), None)
@@ -2027,10 +2004,6 @@ mod tests {
            &ConcreteDataType::date_datatype(),
            &Value::Date(Date::new(1)),
        );
-        check_type_and_value(
-            &ConcreteDataType::datetime_datatype(),
-            &Value::DateTime(DateTime::new(1)),
-        );
        check_type_and_value(
            &ConcreteDataType::timestamp_millisecond_datatype(),
            &Value::Timestamp(Timestamp::new_millisecond(1)),
@@ -2169,11 +2142,6 @@ mod tests {
            serde_json::Value::Number(5000i32.into()),
            to_json(Value::Date(Date::new(5000)))
        );
-        assert_eq!(
-            serde_json::Value::Number(5000i64.into()),
-            to_json(Value::DateTime(DateTime::new(5000)))
-        );
-
        assert_eq!(
            serde_json::Value::Number(1.into()),
            to_json(Value::Timestamp(Timestamp::new_millisecond(1)))
@@ -2259,7 +2227,6 @@ mod tests {
        );

        check_as_value_ref!(Date, Date::new(103));
-        check_as_value_ref!(DateTime, DateTime::new(1034));

        let list = ListValue {
            items: vec![],
@@ -2291,7 +2258,6 @@ mod tests {
        check_as_null!(as_string);
        check_as_null!(as_boolean);
        check_as_null!(as_date);
-        check_as_null!(as_datetime);
        check_as_null!(as_list);

        macro_rules! check_as_correct {
@@ -2304,7 +2270,6 @@ mod tests {
        check_as_correct!("hello".as_bytes(), Binary, as_binary);
        check_as_correct!(true, Boolean, as_boolean);
        check_as_correct!(Date::new(123), Date, as_date);
-        check_as_correct!(DateTime::new(12), DateTime, as_datetime);
        check_as_correct!(Time::new_second(12), Time, as_time);
        check_as_correct!(Duration::new_second(12), Duration, as_duration);
        let list = ListValue {
@@ -2318,7 +2283,6 @@ mod tests {
        assert!(wrong_value.as_string().is_err());
        assert!(wrong_value.as_boolean().is_err());
        assert!(wrong_value.as_date().is_err());
-        assert!(wrong_value.as_datetime().is_err());
        assert!(wrong_value.as_list().is_err());
        assert!(wrong_value.as_time().is_err());
        assert!(wrong_value.as_timestamp().is_err());
@@ -2346,10 +2310,6 @@ mod tests {
            "010203"
        );
        assert_eq!(Value::Date(Date::new(0)).to_string(), "1970-01-01");
-        assert_eq!(
-            Value::DateTime(DateTime::new(0)).to_string(),
-            "1970-01-01 08:00:00+0800"
-        );
        assert_eq!(
            Value::Timestamp(Timestamp::new(1000, TimeUnit::Millisecond)).to_string(),
            "1970-01-01 08:00:01+0800"
@@ -2755,7 +2715,6 @@ mod tests {
        check_value_ref_size_eq(&ValueRef::String("greptimedb"), 10);
        check_value_ref_size_eq(&ValueRef::Binary(b"greptimedb"), 10);
        check_value_ref_size_eq(&ValueRef::Date(Date::new(1)), 4);
-        check_value_ref_size_eq(&ValueRef::DateTime(DateTime::new(1)), 8);
        check_value_ref_size_eq(&ValueRef::Timestamp(Timestamp::new_millisecond(1)), 16);
        check_value_ref_size_eq(&ValueRef::Time(Time::new_millisecond(1)), 16);
        check_value_ref_size_eq(&ValueRef::IntervalYearMonth(IntervalYearMonth::new(1)), 4);
--- a/src/datatypes/src/vectors.rs
+++ b/src/datatypes/src/vectors.rs
@@ -29,7 +29,6 @@ mod binary;
 mod boolean;
 mod constant;
 mod date;
-mod datetime;
 mod decimal;
 mod duration;
 mod eq;
@@ -48,7 +47,6 @@ pub use binary::{BinaryVector, BinaryVectorBuilder};
 pub use boolean::{BooleanVector, BooleanVectorBuilder};
 pub use constant::ConstantVector;
 pub use date::{DateVector, DateVectorBuilder};
-pub use datetime::{DateTimeVector, DateTimeVectorBuilder};
 pub use decimal::{Decimal128Vector, Decimal128VectorBuilder};
 pub use duration::{
    DurationMicrosecondVector, DurationMicrosecondVectorBuilder, DurationMillisecondVector,
@@ -377,7 +375,7 @@ pub mod tests {
        // Test Primitive types
        mutable_primitive_data_type_eq_with_lower!(
            Boolean, Int8, Int16, Int32, Int64, UInt8, UInt16, UInt32, UInt64, Float32, Float64,
-            Date, DateTime, Binary, String
+            Date, Binary, String
        );

        // Test types about time
--- a/src/datatypes/src/vectors/datetime.rs
+++ b/src/datatypes/src/vectors/datetime.rs
@@ -1,116 +0,0 @@
-// 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::types::DateTimeType;
-use crate::vectors::{PrimitiveVector, PrimitiveVectorBuilder};
-
-/// Vector of [`DateTime`](common_time::Date)
-pub type DateTimeVector = PrimitiveVector<DateTimeType>;
-/// Builder for [`DateTimeVector`].
-pub type DateTimeVectorBuilder = PrimitiveVectorBuilder<DateTimeType>;
-
-#[cfg(test)]
-mod tests {
-    use std::sync::Arc;
-
-    use arrow::array::{Array, PrimitiveArray};
-    use arrow_array::ArrayRef;
-    use common_time::timezone::set_default_timezone;
-    use common_time::DateTime;
-
-    use super::*;
-    use crate::data_type::DataType;
-    use crate::prelude::{
-        ConcreteDataType, ScalarVector, ScalarVectorBuilder, Value, ValueRef, Vector, VectorRef,
-    };
-    use crate::serialize::Serializable;
-
-    #[test]
-    fn test_datetime_vector() {
-        set_default_timezone(Some("Asia/Shanghai")).unwrap();
-        let v = DateTimeVector::new(PrimitiveArray::from(vec![1000, 2000, 3000]));
-        assert_eq!(ConcreteDataType::datetime_datatype(), v.data_type());
-        assert_eq!(3, v.len());
-        assert_eq!("DateTimeVector", v.vector_type_name());
-        assert_eq!(
-            &arrow::datatypes::DataType::Date64,
-            v.to_arrow_array().data_type()
-        );
-
-        assert_eq!(Some(DateTime::new(1000)), v.get_data(0));
-        assert_eq!(Value::DateTime(DateTime::new(1000)), v.get(0));
-        assert_eq!(ValueRef::DateTime(DateTime::new(1000)), v.get_ref(0));
-
-        let mut iter = v.iter_data();
-        assert_eq!(Some(DateTime::new(1000)), iter.next().unwrap());
-        assert_eq!(Some(DateTime::new(2000)), iter.next().unwrap());
-        assert_eq!(Some(DateTime::new(3000)), iter.next().unwrap());
-        assert!(!v.is_null(0));
-        assert_eq!(24, v.memory_size());
-
-        if let Value::DateTime(d) = v.get(0) {
-            assert_eq!(1000, d.val());
-        } else {
-            unreachable!()
-        }
-        assert_eq!(
-            "[\"1970-01-01 08:00:01+0800\",\"1970-01-01 08:00:02+0800\",\"1970-01-01 08:00:03+0800\"]",
-            serde_json::to_string(&v.serialize_to_json().unwrap()).unwrap()
-        );
-    }
-
-    #[test]
-    fn test_datetime_vector_builder() {
-        let mut builder = DateTimeVectorBuilder::with_capacity(3);
-        builder.push(Some(DateTime::new(1)));
-        builder.push(None);
-        builder.push(Some(DateTime::new(-1)));
-
-        let v = builder.finish();
-        assert_eq!(ConcreteDataType::datetime_datatype(), v.data_type());
-        assert_eq!(Value::DateTime(DateTime::new(1)), v.get(0));
-        assert_eq!(Value::Null, v.get(1));
-        assert_eq!(Value::DateTime(DateTime::new(-1)), v.get(2));
-
-        let input = DateTimeVector::from_wrapper_slice([
-            DateTime::new(1),
-            DateTime::new(2),
-            DateTime::new(3),
-        ]);
-
-        let mut builder = DateTimeType.create_mutable_vector(3);
-        builder.push_value_ref(ValueRef::DateTime(DateTime::new(5)));
-        assert!(builder.try_push_value_ref(ValueRef::Int32(123)).is_err());
-        builder.extend_slice_of(&input, 1, 2).unwrap();
-        assert!(builder
-            .extend_slice_of(&crate::vectors::Int32Vector::from_slice([13]), 0, 1)
-            .is_err());
-        let vector = builder.to_vector();
-
-        let expect: VectorRef = Arc::new(DateTimeVector::from_wrapper_slice([
-            DateTime::new(5),
-            DateTime::new(2),
-            DateTime::new(3),
-        ]));
-        assert_eq!(expect, vector);
-    }
-
-    #[test]
-    fn test_datetime_from_arrow() {
-        let vector = DateTimeVector::from_wrapper_slice([DateTime::new(1), DateTime::new(2)]);
-        let arrow: ArrayRef = Arc::new(vector.as_arrow().slice(0, vector.len())) as _;
-        let vector2 = DateTimeVector::try_from_arrow_array(arrow).unwrap();
-        assert_eq!(vector, vector2);
-    }
-}
--- a/src/datatypes/src/vectors/eq.rs
+++ b/src/datatypes/src/vectors/eq.rs
@@ -20,12 +20,12 @@ use crate::data_type::DataType;
 use crate::types::{DurationType, TimeType, TimestampType};
 use crate::vectors::constant::ConstantVector;
 use crate::vectors::{
-    BinaryVector, BooleanVector, DateTimeVector, DateVector, Decimal128Vector,
-    DurationMicrosecondVector, DurationMillisecondVector, DurationNanosecondVector,
-    DurationSecondVector, IntervalDayTimeVector, IntervalMonthDayNanoVector,
-    IntervalYearMonthVector, ListVector, PrimitiveVector, StringVector, TimeMicrosecondVector,
-    TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector,
-    TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, Vector,
+    BinaryVector, BooleanVector, DateVector, Decimal128Vector, DurationMicrosecondVector,
+    DurationMillisecondVector, DurationNanosecondVector, DurationSecondVector,
+    IntervalDayTimeVector, IntervalMonthDayNanoVector, IntervalYearMonthVector, ListVector,
+    PrimitiveVector, StringVector, TimeMicrosecondVector, TimeMillisecondVector,
+    TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector, TimestampMillisecondVector,
+    TimestampNanosecondVector, TimestampSecondVector, Vector,
 };
 use crate::with_match_primitive_type_id;

@@ -83,7 +83,6 @@ fn equal(lhs: &dyn Vector, rhs: &dyn Vector) -> bool {
        Binary(_) | Json(_) | Vector(_) => is_vector_eq!(BinaryVector, lhs, rhs),
        String(_) => is_vector_eq!(StringVector, lhs, rhs),
        Date(_) => is_vector_eq!(DateVector, lhs, rhs),
-        DateTime(_) => is_vector_eq!(DateTimeVector, lhs, rhs),
        Timestamp(t) => match t {
            TimestampType::Second(_) => {
                is_vector_eq!(TimestampSecondVector, lhs, rhs)
@@ -195,7 +194,6 @@ mod tests {
        )));
        assert_vector_ref_eq(Arc::new(BooleanVector::from(vec![true, false])));
        assert_vector_ref_eq(Arc::new(DateVector::from(vec![Some(100), Some(120)])));
-        assert_vector_ref_eq(Arc::new(DateTimeVector::from(vec![Some(100), Some(120)])));
        assert_vector_ref_eq(Arc::new(TimestampSecondVector::from_values([100, 120])));
        assert_vector_ref_eq(Arc::new(TimestampMillisecondVector::from_values([
            100, 120,
--- a/src/datatypes/src/vectors/helper.rs
+++ b/src/datatypes/src/vectors/helper.rs
@@ -31,7 +31,7 @@ use crate::prelude::DataType;
 use crate::scalars::{Scalar, ScalarVectorBuilder};
 use crate::value::{ListValue, ListValueRef, Value};
 use crate::vectors::{
-    BinaryVector, BooleanVector, ConstantVector, DateTimeVector, DateVector, Decimal128Vector,
+    BinaryVector, BooleanVector, ConstantVector, DateVector, Decimal128Vector,
    DurationMicrosecondVector, DurationMillisecondVector, DurationNanosecondVector,
    DurationSecondVector, Float32Vector, Float64Vector, Int16Vector, Int32Vector, Int64Vector,
    Int8Vector, IntervalDayTimeVector, IntervalMonthDayNanoVector, IntervalYearMonthVector,
@@ -179,9 +179,6 @@ impl Helper {
            ScalarValue::Date32(v) => {
                ConstantVector::new(Arc::new(DateVector::from(vec![v])), length)
            }
-            ScalarValue::Date64(v) => {
-                ConstantVector::new(Arc::new(DateTimeVector::from(vec![v])), length)
-            }
            ScalarValue::TimestampSecond(v, _) => {
                // Timezone is unimplemented now.
                ConstantVector::new(Arc::new(TimestampSecondVector::from(vec![v])), length)
@@ -244,7 +241,8 @@ impl Helper {
            | ScalarValue::Float16(_)
            | ScalarValue::Utf8View(_)
            | ScalarValue::BinaryView(_)
-            | ScalarValue::Map(_) => {
+            | ScalarValue::Map(_)
+            | ScalarValue::Date64(_) => {
                return error::ConversionSnafu {
                    from: format!("Unsupported scalar value: {value}"),
                }
@@ -286,7 +284,6 @@ impl Helper {
                Arc::new(StringVector::try_from_arrow_array(array)?)
            }
            ArrowDataType::Date32 => Arc::new(DateVector::try_from_arrow_array(array)?),
-            ArrowDataType::Date64 => Arc::new(DateTimeVector::try_from_arrow_array(array)?),
            ArrowDataType::List(_) => Arc::new(ListVector::try_from_arrow_array(array)?),
            ArrowDataType::Timestamp(unit, _) => match unit {
                TimeUnit::Second => Arc::new(TimestampSecondVector::try_from_arrow_array(array)?),
@@ -362,7 +359,8 @@ impl Helper {
            | ArrowDataType::BinaryView
            | ArrowDataType::Utf8View
            | ArrowDataType::ListView(_)
-            | ArrowDataType::LargeListView(_) => {
+            | ArrowDataType::LargeListView(_)
+            | ArrowDataType::Date64 => {
                return error::UnsupportedArrowTypeSnafu {
                    arrow_type: array.as_ref().data_type().clone(),
                }
@@ -411,9 +409,9 @@ impl Helper {
 #[cfg(test)]
 mod tests {
    use arrow::array::{
-        ArrayRef, BooleanArray, Date32Array, Date64Array, Float32Array, Float64Array, Int16Array,
-        Int32Array, Int64Array, Int8Array, LargeBinaryArray, ListArray, NullArray,
-        Time32MillisecondArray, Time32SecondArray, Time64MicrosecondArray, Time64NanosecondArray,
+        ArrayRef, BooleanArray, Date32Array, Float32Array, Float64Array, Int16Array, Int32Array,
+        Int64Array, Int8Array, LargeBinaryArray, ListArray, NullArray, Time32MillisecondArray,
+        Time32SecondArray, Time64MicrosecondArray, Time64NanosecondArray,
        TimestampMicrosecondArray, TimestampMillisecondArray, TimestampNanosecondArray,
        TimestampSecondArray, UInt16Array, UInt32Array, UInt64Array, UInt8Array,
    };
@@ -424,7 +422,7 @@ mod tests {
    use common_decimal::Decimal128;
    use common_time::time::Time;
    use common_time::timestamp::TimeUnit;
-    use common_time::{Date, DateTime, Duration};
+    use common_time::{Date, Duration};

    use super::*;
    use crate::value::Value;
@@ -466,16 +464,6 @@ mod tests {
        }
    }

-    #[test]
-    fn test_try_from_scalar_datetime_value() {
-        let vector = Helper::try_from_scalar_value(ScalarValue::Date64(Some(42)), 3).unwrap();
-        assert_eq!(ConcreteDataType::datetime_datatype(), vector.data_type());
-        assert_eq!(3, vector.len());
-        for i in 0..vector.len() {
-            assert_eq!(Value::DateTime(DateTime::new(42)), vector.get(i));
-        }
-    }
-
    #[test]
    fn test_try_from_scalar_duration_value() {
        let vector =
@@ -606,7 +594,6 @@ mod tests {
        check_try_into_vector(Float64Array::from(vec![1.0, 2.0, 3.0]));
        check_try_into_vector(StringArray::from(vec!["hello", "world"]));
        check_try_into_vector(Date32Array::from(vec![1, 2, 3]));
-        check_try_into_vector(Date64Array::from(vec![1, 2, 3]));
        let data = vec![None, Some(vec![Some(6), Some(7)])];
        let list_array = ListArray::from_iter_primitive::<Int32Type, _, _>(data);
        check_try_into_vector(list_array);
@@ -734,7 +721,6 @@ mod tests {
        check_into_and_from(Float64Array::from(vec![1.0, 2.0, 3.0]));
        check_into_and_from(StringArray::from(vec!["hello", "world"]));
        check_into_and_from(Date32Array::from(vec![1, 2, 3]));
-        check_into_and_from(Date64Array::from(vec![1, 2, 3]));

        check_into_and_from(TimestampSecondArray::from(vec![1, 2, 3]));
        check_into_and_from(TimestampMillisecondArray::from(vec![1, 2, 3]));
--- a/src/datatypes/src/vectors/operations/filter.rs
+++ b/src/datatypes/src/vectors/operations/filter.rs
@@ -32,7 +32,7 @@ pub(crate) use filter_non_constant;
 mod tests {
    use std::sync::Arc;

-    use common_time::{Date, DateTime};
+    use common_time::Date;

    use crate::scalars::ScalarVector;
    use crate::timestamp::{
@@ -127,8 +127,6 @@ mod tests {
    #[test]
    fn test_filter_date_like() {
        impl_filter_date_like_test!(DateVector, Date, new);
-        impl_filter_date_like_test!(DateTimeVector, DateTime, new);
-
        impl_filter_date_like_test!(TimestampSecondVector, TimestampSecond, from_native);
        impl_filter_date_like_test!(
            TimestampMillisecondVector,
--- a/src/datatypes/src/vectors/operations/find_unique.rs
+++ b/src/datatypes/src/vectors/operations/find_unique.rs
@@ -105,7 +105,7 @@ pub(crate) fn find_unique_constant(
 mod tests {
    use std::sync::Arc;

-    use common_time::{Date, DateTime};
+    use common_time::Date;

    use super::*;
    use crate::timestamp::*;
@@ -358,7 +358,6 @@ mod tests {
    #[test]
    fn test_find_unique_date_like() {
        impl_find_unique_date_like_test!(DateVector, Date, new);
-        impl_find_unique_date_like_test!(DateTimeVector, DateTime, new);
        impl_find_unique_date_like_test!(TimestampSecondVector, TimestampSecond, from);
        impl_find_unique_date_like_test!(TimestampMillisecondVector, TimestampMillisecond, from);
        impl_find_unique_date_like_test!(TimestampMicrosecondVector, TimestampMicrosecond, from);
--- a/src/datatypes/src/vectors/operations/replicate.rs
+++ b/src/datatypes/src/vectors/operations/replicate.rs
@@ -41,7 +41,7 @@ mod tests {
    use std::sync::Arc;

    use common_time::timestamp::TimeUnit;
-    use common_time::{Date, DateTime, Timestamp};
+    use common_time::{Date, Timestamp};
    use paste::paste;

    use super::*;
@@ -161,8 +161,6 @@ mod tests {
    #[test]
    fn test_replicate_date_like() {
        impl_replicate_date_like_test!(DateVector, Date, new);
-        impl_replicate_date_like_test!(DateTimeVector, DateTime, new);
-
        impl_replicate_timestamp_test!(Second);
        impl_replicate_timestamp_test!(Millisecond);
        impl_replicate_timestamp_test!(Microsecond);
--- a/src/datatypes/src/vectors/operations/take.rs
+++ b/src/datatypes/src/vectors/operations/take.rs
@@ -33,7 +33,7 @@ mod tests {
    use std::sync::Arc;

    use arrow::array::{PrimitiveArray, UInt32Array};
-    use common_time::{Date, DateTime};
+    use common_time::Date;

    use crate::prelude::VectorRef;
    use crate::scalars::ScalarVector;
@@ -105,7 +105,6 @@ mod tests {

        // test date like type
        take_time_like_test!(DateVector, Date, new);
-        take_time_like_test!(DateTimeVector, DateTime, new);
        take_time_like_test!(TimestampSecondVector, TimestampSecond, from_native);
        take_time_like_test!(
            TimestampMillisecondVector,
--- a/src/flow/Cargo.toml
+++ b/src/flow/Cargo.toml
@@ -46,6 +46,7 @@ get-size2 = "0.1.2"
 greptime-proto.workspace = true
 # This fork of hydroflow is simply for keeping our dependency in our org, and pin the version
 # otherwise it is the same with upstream repo
+chrono.workspace = true
 http.workspace = true
 hydroflow = { git = "https://github.com/GreptimeTeam/hydroflow.git", branch = "main" }
 itertools.workspace = true
--- a/src/flow/src/adapter/flownode_impl.rs
+++ b/src/flow/src/adapter/flownode_impl.rs
@@ -13,7 +13,6 @@
 // limitations under the License.

 //! impl `FlowNode` trait for FlowNodeManager so standalone can call them
-
 use std::collections::HashMap;

 use api::v1::flow::{
--- a/src/flow/src/adapter/stat.rs
+++ b/src/flow/src/adapter/stat.rs
@@ -21,19 +21,31 @@ use crate::FlowWorkerManager;
 impl FlowWorkerManager {
    pub async fn gen_state_report(&self) -> FlowStat {
        let mut full_report = BTreeMap::new();
+        let mut last_exec_time_map = BTreeMap::new();
        for worker in self.worker_handles.iter() {
            match worker.get_state_size().await {
                Ok(state_size) => {
-                    full_report.extend(state_size.into_iter().map(|(k, v)| (k as u32, v)))
+                    full_report.extend(state_size.into_iter().map(|(k, v)| (k as u32, v)));
                }
                Err(err) => {
                    common_telemetry::error!(err; "Get flow stat size error");
                }
            }
+
+            match worker.get_last_exec_time_map().await {
+                Ok(last_exec_time) => {
+                    last_exec_time_map
+                        .extend(last_exec_time.into_iter().map(|(k, v)| (k as u32, v)));
+                }
+                Err(err) => {
+                    common_telemetry::error!(err; "Get last exec time error");
+                }
+            }
        }

        FlowStat {
            state_size: full_report,
+            last_exec_time_map,
        }
    }
 }
--- a/src/flow/src/adapter/tests.rs
+++ b/src/flow/src/adapter/tests.rs
@@ -41,7 +41,7 @@ pub fn new_test_table_info_with_name<I: IntoIterator<Item = u32>>(
        .build()
        .unwrap();

-    let meta = TableMetaBuilder::default()
+    let meta = TableMetaBuilder::empty()
        .schema(Arc::new(schema))
        .primary_key_indices(vec![0])
        .engine("engine")
--- a/src/flow/src/adapter/worker.rs
+++ b/src/flow/src/adapter/worker.rs
@@ -98,6 +98,10 @@ impl<'subgraph> ActiveDataflowState<'subgraph> {
        self.state.set_current_ts(ts);
    }

+    pub fn set_last_exec_time(&mut self, ts: repr::Timestamp) {
+        self.state.set_last_exec_time(ts);
+    }
+
    /// Run all available subgraph
    ///
    /// return true if any subgraph actually executed
@@ -212,6 +216,21 @@ impl WorkerHandle {
            .build()
        })
    }
+
+    pub async fn get_last_exec_time_map(&self) -> Result<BTreeMap<FlowId, i64>, Error> {
+        let ret = self
+            .itc_client
+            .call_with_resp(Request::QueryLastExecTimeMap)
+            .await?;
+        ret.into_query_last_exec_time_map().map_err(|ret| {
+            InternalSnafu {
+                reason: format!(
+                    "Flow Node/Worker get_last_exec_time_map failed, expect Response::QueryLastExecTimeMap, found {ret:?}"
+                ),
+            }
+            .build()
+        })
+    }
 }

 impl Drop for WorkerHandle {
@@ -335,6 +354,7 @@ impl<'s> Worker<'s> {
    pub fn run_tick(&mut self, now: repr::Timestamp) {
        for (_flow_id, task_state) in self.task_states.iter_mut() {
            task_state.set_current_ts(now);
+            task_state.set_last_exec_time(now);
            task_state.run_available();
        }
    }
@@ -395,6 +415,15 @@ impl<'s> Worker<'s> {
                }
                Some(Response::QueryStateSize { result: ret })
            }
+            Request::QueryLastExecTimeMap => {
+                let mut ret = BTreeMap::new();
+                for (flow_id, task_state) in self.task_states.iter() {
+                    if let Some(last_exec_time) = task_state.state.last_exec_time() {
+                        ret.insert(*flow_id, last_exec_time);
+                    }
+                }
+                Some(Response::QueryLastExecTimeMap { result: ret })
+            }
        };
        Ok(ret)
    }
@@ -427,6 +456,7 @@ pub enum Request {
    },
    Shutdown,
    QueryStateSize,
+    QueryLastExecTimeMap,
 }

 #[derive(Debug, EnumAsInner)]
@@ -446,6 +476,10 @@ enum Response {
        /// each flow tasks' state size
        result: BTreeMap<FlowId, usize>,
    },
+    QueryLastExecTimeMap {
+        /// each flow tasks' last execution time
+        result: BTreeMap<FlowId, i64>,
+    },
 }

 fn create_inter_thread_call() -> (InterThreadCallClient, InterThreadCallServer) {
--- a/src/flow/src/compute/render/map.rs
+++ b/src/flow/src/compute/render/map.rs
@@ -290,7 +290,9 @@ mod test {
        let mfp = MapFilterProject::new(1)
            .filter(vec![
                ScalarExpr::Column(0)
-                    .call_unary(expr::UnaryFunc::Cast(ConcreteDataType::datetime_datatype()))
+                    .call_unary(expr::UnaryFunc::Cast(
+                        ConcreteDataType::timestamp_microsecond_datatype(),
+                    ))
                    .call_binary(
                        ScalarExpr::CallUnmaterializable(expr::UnmaterializableFunc::Now),
                        BinaryFunc::Gte,
@@ -300,7 +302,9 @@ mod test {
                        ScalarExpr::literal(4i64.into(), ConcreteDataType::int64_datatype()),
                        BinaryFunc::SubInt64,
                    )
-                    .call_unary(expr::UnaryFunc::Cast(ConcreteDataType::datetime_datatype()))
+                    .call_unary(expr::UnaryFunc::Cast(
+                        ConcreteDataType::timestamp_microsecond_datatype(),
+                    ))
                    .call_binary(
                        ScalarExpr::CallUnmaterializable(expr::UnmaterializableFunc::Now),
                        BinaryFunc::Lt,
--- a/src/flow/src/compute/state.rs
+++ b/src/flow/src/compute/state.rs
@@ -45,6 +45,8 @@ pub struct DataflowState {
    arrange_used: Vec<ArrangeHandler>,
    /// the time arrangement need to be expired after a certain time in milliseconds
    expire_after: Option<Timestamp>,
+    /// the last time each subgraph executed
+    last_exec_time: Option<Timestamp>,
 }

 impl DataflowState {
@@ -114,6 +116,14 @@ impl DataflowState {
    pub fn get_state_size(&self) -> usize {
        self.arrange_used.iter().map(|x| x.read().get_size()).sum()
    }
+
+    pub fn set_last_exec_time(&mut self, time: Timestamp) {
+        self.last_exec_time = Some(time);
+    }
+
+    pub fn last_exec_time(&self) -> Option<Timestamp> {
+        self.last_exec_time
+    }
 }

 #[derive(Debug, Clone)]
--- a/src/flow/src/df_optimizer.rs
+++ b/src/flow/src/df_optimizer.rs
@@ -479,7 +479,7 @@ impl ScalarUDFImpl for TumbleExpand {
        match (arg_types.first(), arg_types.get(1), arg_types.get(2)) {
            (Some(ts), Some(window), opt) => {
                use arrow_schema::DataType::*;
-                if !matches!(ts, Date32 | Date64 | Timestamp(_, _)) {
+                if !matches!(ts, Date32 | Timestamp(_, _)) {
                    return Err(DataFusionError::Plan(
                        format!("Expect timestamp column as first arg for tumble_start, found {:?}", ts)
                    ));
@@ -491,7 +491,7 @@ impl ScalarUDFImpl for TumbleExpand {
                }

                if let Some(start_time) = opt{
-                    if !matches!(start_time,  Utf8 | Date32 | Date64 | Timestamp(_, _)){
+                    if !matches!(start_time,  Utf8 | Date32 | Timestamp(_, _)){
                        return Err(DataFusionError::Plan(
                            format!("Expect start_time to either be date, timestamp or string, found {:?}", start_time)
                        ));
--- a/src/flow/src/expr/func.rs
+++ b/src/flow/src/expr/func.rs
@@ -554,8 +554,6 @@ fn get_ts_as_millisecond(arg: Value) -> Result<repr::Timestamp, EvalError> {
        ts.convert_to(TimeUnit::Millisecond)
            .context(OverflowSnafu)?
            .value()
-    } else if let Some(ts) = arg.as_datetime() {
-        ts.val()
    } else {
        InvalidArgumentSnafu {
            reason: "Expect input to be timestamp or datetime type",
--- a/src/flow/src/expr/relation/accum.rs
+++ b/src/flow/src/expr/relation/accum.rs
@@ -759,7 +759,7 @@ fn ty_eq_without_precision(left: ConcreteDataType, right: ConcreteDataType) -> b
 #[allow(clippy::too_many_lines)]
 #[cfg(test)]
 mod test {
-    use common_time::DateTime;
+    use common_time::Timestamp;

    use super::*;

@@ -813,13 +813,13 @@ mod test {
            (
                AggregateFunc::MaxDateTime,
                vec![
-                    (Value::DateTime(DateTime::from(0)), 1),
-                    (Value::DateTime(DateTime::from(1)), 1),
+                    (Value::Timestamp(Timestamp::from(0)), 1),
+                    (Value::Timestamp(Timestamp::from(1)), 1),
                    (Value::Null, 1),
                ],
                (
-                    Value::DateTime(DateTime::from(1)),
-                    vec![Value::DateTime(DateTime::from(1)), 2i64.into()],
+                    Value::Timestamp(Timestamp::from(1)),
+                    vec![Value::Timestamp(Timestamp::from(1)), 2i64.into()],
                ),
            ),
            (
--- a/src/flow/src/expr/relation/func.rs
+++ b/src/flow/src/expr/relation/func.rs
@@ -267,7 +267,7 @@ impl AggregateFunc {
            MaxBool => (boolean_datatype, Max),
            MaxString => (string_datatype, Max),
            MaxDate => (date_datatype, Max),
-            MaxDateTime => (datetime_datatype, Max),
+            MaxDateTime => (timestamp_microsecond_datatype, Max),
            MaxTimestamp => (timestamp_second_datatype, Max),
            MaxTime => (time_second_datatype, Max),
            MaxDuration => (duration_second_datatype, Max),
@@ -283,7 +283,7 @@ impl AggregateFunc {
            MinBool => (boolean_datatype, Min),
            MinString => (string_datatype, Min),
            MinDate => (date_datatype, Min),
-            MinDateTime => (datetime_datatype, Min),
+            MinDateTime => (timestamp_microsecond_datatype, Min),
            MinTimestamp => (timestamp_second_datatype, Min),
            MinTime => (time_second_datatype, Min),
            MinDuration => (duration_second_datatype, Min),
--- a/src/flow/src/heartbeat.rs
+++ b/src/flow/src/heartbeat.rs
@@ -154,14 +154,18 @@ impl HeartbeatTask {
        };
        let flow_stat = latest_report
            .as_ref()
-            .map(|report| {
-                report
+            .map(|report| api::v1::meta::FlowStat {
+                flow_stat_size: report
                    .state_size
                    .iter()
                    .map(|(k, v)| (*k, *v as u64))
-                    .collect()
-            })
-            .map(|f| api::v1::meta::FlowStat { flow_stat_size: f });
+                    .collect(),
+                flow_last_exec_time_map: report
+                    .last_exec_time_map
+                    .iter()
+                    .map(|(k, v)| (*k, *v))
+                    .collect(),
+            });

        Some(HeartbeatRequest {
            mailbox_message,
--- a/src/flow/src/repr.rs
+++ b/src/flow/src/repr.rs
@@ -68,13 +68,10 @@ pub fn value_to_internal_ts(value: Value) -> Result<i64, EvalError> {
        let ty = arg.data_type();
        matches!(
            ty,
-            ConcreteDataType::Date(..)
-                | ConcreteDataType::DateTime(..)
-                | ConcreteDataType::Timestamp(..)
+            ConcreteDataType::Date(..) | ConcreteDataType::Timestamp(..)
        )
    };
    match value {
-        Value::DateTime(ts) => Ok(ts.val()),
        Value::Int64(ts) => Ok(ts),
        arg if is_supported_time_type(&arg) => {
            let arg_ty = arg.data_type();
@@ -214,7 +211,7 @@ impl From<Row> for ProtoRow {
 }
 #[cfg(test)]
 mod test {
-    use common_time::{Date, DateTime};
+    use common_time::{Date, Timestamp};

    use super::*;

@@ -244,7 +241,7 @@ mod test {
        {
            let a = Value::from(1i32);
            let b = Value::from(1i64);
-            let c = Value::DateTime(DateTime::new(1i64));
+            let c = Value::Timestamp(Timestamp::new_millisecond(1i64));
            let d = Value::from(1.0);

            assert!(value_to_internal_ts(a).is_err());
--- a/src/frontend/src/instance/builder.rs
+++ b/src/frontend/src/instance/builder.rs
@@ -133,6 +133,7 @@ impl FrontendBuilder {
                .context(error::CacheRequiredSnafu {
                    name: TABLE_FLOWNODE_SET_CACHE_NAME,
                })?;
+
        let inserter = Arc::new(Inserter::new(
            self.catalog_manager.clone(),
            partition_manager.clone(),
--- a/src/meta-srv/Cargo.toml
+++ b/src/meta-srv/Cargo.toml
@@ -61,6 +61,7 @@ serde_json.workspace = true
 servers.workspace = true
 snafu.workspace = true
 store-api.workspace = true
+strum.workspace = true
 table.workspace = true
 tokio.workspace = true
 tokio-postgres = { workspace = true, optional = true, features = ["with-chrono-0_4"] }
--- a/src/meta-srv/src/error.rs
+++ b/src/meta-srv/src/error.rs
@@ -749,6 +749,41 @@ pub enum Error {
        location: Location,
        source: common_meta::error::Error,
    },
+
+    #[snafu(display("Logical table cannot add follower: {table_id}"))]
+    LogicalTableCannotAddFollower {
+        table_id: TableId,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display(
+        "A region follower cannot be placed on the same node as the leader: {region_id}, {peer_id}"
+    ))]
+    RegionFollowerLeaderConflict {
+        region_id: RegionId,
+        peer_id: u64,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display(
+        "Multiple region followers cannot be placed on the same node: {region_id}, {peer_id}"
+    ))]
+    MultipleRegionFollowersOnSameNode {
+        region_id: RegionId,
+        peer_id: u64,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
+    #[snafu(display("Region follower not exists: {region_id}, {peer_id}"))]
+    RegionFollowerNotExists {
+        region_id: RegionId,
+        peer_id: u64,
+        #[snafu(implicit)]
+        location: Location,
+    },
 }

 impl Error {
@@ -818,6 +853,10 @@ impl ErrorExt for Error {
            | Error::ProcedureNotFound { .. }
            | Error::TooManyPartitions { .. }
            | Error::TomlFormat { .. }
+            | Error::LogicalTableCannotAddFollower { .. }
+            | Error::RegionFollowerLeaderConflict { .. }
+            | Error::MultipleRegionFollowersOnSameNode { .. }
+            | Error::RegionFollowerNotExists { .. }
            | Error::HandlerNotFound { .. } => StatusCode::InvalidArguments,
            Error::LeaseKeyFromUtf8 { .. }
            | Error::LeaseValueFromUtf8 { .. }
--- a/src/meta-srv/src/handler/flow_state_handler.rs
+++ b/src/meta-srv/src/handler/flow_state_handler.rs
@@ -42,12 +42,20 @@ impl HeartbeatHandler for FlowStateHandler {
        _ctx: &mut Context,
        _acc: &mut HeartbeatAccumulator,
    ) -> Result<HandleControl> {
-        if let Some(FlowStat { flow_stat_size }) = &req.flow_stat {
+        if let Some(FlowStat {
+            flow_stat_size,
+            flow_last_exec_time_map,
+        }) = &req.flow_stat
+        {
            let state_size = flow_stat_size
                .iter()
                .map(|(k, v)| (*k, *v as usize))
                .collect();
-            let value = FlowStateValue::new(state_size);
+            let last_exec_time_map = flow_last_exec_time_map
+                .iter()
+                .map(|(k, v)| (*k, *v))
+                .collect();
+            let value: FlowStateValue = FlowStateValue::new(state_size, last_exec_time_map);
            self.flow_state_manager
                .put(value)
                .await
--- a/src/meta-srv/src/metasrv.rs
+++ b/src/meta-srv/src/metasrv.rs
@@ -70,7 +70,7 @@ use crate::state::{become_follower, become_leader, StateRef};

 pub const TABLE_ID_SEQ: &str = "table_id";
 pub const FLOW_ID_SEQ: &str = "flow_id";
-pub const METASRV_HOME: &str = "/tmp/metasrv";
+pub const METASRV_HOME: &str = "./greptimedb_data/metasrv";

 #[cfg(feature = "pg_kvbackend")]
 pub const DEFAULT_META_TABLE_NAME: &str = "greptime_metakv";
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
evenyag	ab1928d5fd	feat: refetch sst on open	2025-03-17 15:17:14 +08:00
Sicong Hu	09dacc8e9b	feat: add `vec_subvector` function (#5683 ) * feat: add vec_subvector function * change datatype of arg1 and arg2 from u64 to i64 * add sqlness test * improve description comments	2025-03-16 10:43:53 +00:00
Ruihang Xia	dec439db2b	chore: bump version to 0.14.0 (#5711 ) Signed-off-by: Ruihang Xia <waynestxia@gmail.com>	2025-03-16 09:58:19 +00:00
Ning Sun	dc76571166	feat: move default data path from /tmp to current directory (#5719 )	2025-03-16 09:57:46 +00:00
shuiyisong	3e17f8c426	chore: use `Bytes` instead of string in bulk ingestion (#5717 ) chore: use bytes instead of string in bulk log ingestion	2025-03-14 09:31:35 +00:00
yihong	a5df3954f3	chore: update flate2 version (#5706 ) Signed-off-by: yihong0618 <zouzou0208@gmail.com>	2025-03-14 02:15:27 +00:00
Ruihang Xia	32fd850c20	perf: support in list in simple filter (#5709 ) * feat: support in list in simple filter Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * fix clippy Signed-off-by: Ruihang Xia <waynestxia@gmail.com> --------- Signed-off-by: Ruihang Xia <waynestxia@gmail.com>	2025-03-14 01:08:29 +00:00
shuiyisong	2bfdae4f8f	feat: add simple extract processor (#5688 ) * feat: add simple extract processor * chore: add test * chore: add license header * chore: minor update	2025-03-13 09:19:58 +00:00
shuiyisong	fcb898e9a4	chore: support `inverted` index in pipeline (#5700 ) chore: rebase main	2025-03-13 08:30:29 +00:00
Ning Sun	8fa2fdfc42	feat: make empty parent_span_id null for v1 (#5690 )	2025-03-13 07:48:15 +00:00
shuiyisong	4dc1a1d60f	chore: support `tag` in transform (#5701 ) chore: support tag in transform to specify tag	2025-03-13 07:27:12 +00:00
Lei, HUANG	e375a18011	fix: conversion from TableMeta to TableMetaBuilder (#5693 ) * refactor: use proc macro to generate conversion between TableMeta and TableMetaBuilder * chore: format * fix/partition-key-index: ### Update `TableMeta` and Add Partition and Alter Table Tests - `metadata.rs`: Modified `new_meta_builder` method in `TableMeta` to manually remove `value_indices` by setting it to `None` in the `TableMetaBuilder`. - `partition_and_alter.result` & `partition_and_alter.sql`: Added new test cases for creating, inserting, selecting, altering, and dropping a partitioned table `molestiAe`. These tests verify partitioning on the `sImiLiQUE` column and altering the table with a TTL setting. fix/partition-key-index: ### Remove Obsolete TODO Comment in `metadata.rs` - Removed an outdated TODO comment regarding the `new_meta_builder` function in `src/table/src/metadata.rs`. chore: check struct name in derive_meta_builder refactor: Simplify TableMeta struct name check in macro refactor: Improve ToMetaBuilder derive macro validation and error handling refactor: Enforce ToMetaBuilder macro for table::metadata::TableMeta struct * fix/partition-key-index: Update `partition_and_alter.sql` to modify TTL setting - Modified the TTL setting for the `molestiAe` table to '1d' in `partition_and_alter.sql`. * fix: sqlness * fix/partition-key-index: ### Update `TableMeta` and Test File Structure - Enhancement: Added a note in `metadata.rs` to always use `new_meta_builder` for creating `TableMetaBuilder`. - Refactor: Renamed test result and SQL files for better organization: - `partition_and_alter.result` to `alter/partition_and_alter.result` - `partition_and_alter.sql` to `alter/partition_and_alter.sql` * refactor: Simplify `derive_meta_builder` by initializing fields with `Default::default()` * fix/partition-key-index: ### Commit Summary - Refactor `TableMetaBuilder` Initialization: - Replaced `TableMetaBuilder::default()` with `TableMetaBuilder::empty()` across multiple files for initializing `TableMetaBuilder` instances. - Affected files include: - `src/catalog/src/system_schema.rs` - `src/common/meta/src/key/test_utils.rs` - `src/operator/src/req_convert/insert/fill_impure_default.rs` - `src/query/src/log_query/planner.rs` - `src/query/src/promql/planner.rs` - `src/query/src/range_select/plan_rewrite.rs` - `src/query/src/sql/show_create_table.rs` - `src/table/src/test_util/memtable.rs` - `src/table/src/test_util/table_info.rs` - Enhance `TableMetaBuilder`: - Added `custom_constructor` to `TableMeta` and implemented an `empty` method for `TableMetaBuilder`. - Modified `TableMetaBuilder` to include a `new_external_table` method with default values. - Updated `src/table/src/metadata.rs` to reflect these changes. - Add Testing Feature: - Introduced a conditional compilation for `test_util` in `src/table/src/lib.rs` to include testing utilities when the `testing` feature is enabled. - Update `Cargo.toml`: - Enabled the `testing` feature for the `table` module in `src/common/meta/Cargo.toml`. - Modify `NumbersTable` Initialization: - Replaced `TableMetaBuilder` with direct `TableMeta` struct initialization in `src/table/src/table/numbers.rs`. - Test Result Update: - Updated test results in `tests/cases/standalone/common/alter/partition_and_alter.result` to reflect changes in table meta handling. * fix: rename default to empty * docs: add doc for TableMetaBuilder::empty * chore: Update src/table/src/metadata.rs --------- Co-authored-by: Yingwen <realevenyag@gmail.com>	2025-03-13 06:30:16 +00:00
shuiyisong	e0ff701e51	chore: support `application/x-ndjson` for log ingest (#5697 ) chore: support ndjson content type	2025-03-13 04:29:22 +00:00
Yingwen	25645a3303	feat: expose virtual_host_style config for s3 storage (#5696 ) * feat: expose enable_virtual_host_style for s3 storage * docs: update examples * test: fix config test	2025-03-12 13:46:56 +00:00
Ruihang Xia	b32ea7d84c	feat: add Docker image tag information to step summary in dev-build workflow (#5692 ) Signed-off-by: Ruihang Xia <waynestxia@gmail.com>	2025-03-12 13:45:19 +00:00
discord9	f164f6eaf3	fix: FlowInfoValue's compatibility (#5695 )	2025-03-12 09:02:48 +00:00
Yohan Wal	af1920defc	feat: add mysql kvbackend (#5528 ) * feat: add mysql kvbackend txn support * chore: error handling * chore: follow review comments * chore: follow review comments * chore: follow review comments * revert: mysql QAQ * revert: revert changes to sqls This reverts commit `cf98c50dd9`. * chore: add comments	2025-03-12 06:52:56 +00:00
Lei, HUANG	7c97fae522	chore: check region wal provider on startup to avoid inconsistence (#5687 )	2025-03-11 17:51:18 +00:00
AntiTopQuark	b8070adc3a	feat: enhancement information_schema.flows (#5623 ) * feat: enhancement information_schema.flows * feat: enhancement information_schema.flows * u * u * u * u * u * u * u * u * u * update * update * update * delete unused code * u * u * Update src/flow/src/adapter/worker.rs Co-authored-by: dennis zhuang <killme2008@gmail.com> * Update src/common/meta/src/key/flow/flow_state.rs Co-authored-by: dennis zhuang <killme2008@gmail.com> * Update src/common/meta/src/key/flow/flow_info.rs Co-authored-by: dennis zhuang <killme2008@gmail.com> * Update src/common/meta/src/key/flow/flow_state.rs Co-authored-by: dennis zhuang <killme2008@gmail.com> * Update src/common/meta/src/key/flow/flow_info.rs Co-authored-by: dennis zhuang <killme2008@gmail.com> * u * u * u * u * u * u * chore: fix sqlness * chore: update proto * fix: remove date time * fix: update result of information_schema test --------- Co-authored-by: dennis zhuang <killme2008@gmail.com> Co-authored-by: discord9 <discord9@163.com>	2025-03-11 15:08:10 +00:00
yihong	11bfb17328	feat: support export command export data to s3 (#5585 ) * feat: s3 first step Signed-off-by: yihong0618 <zouzou0208@gmail.com> * fix: finish s3 export Signed-off-by: yihong0618 <zouzou0208@gmail.com> * fix: drop useless comment Signed-off-by: yihong0618 <zouzou0208@gmail.com> * fix: forget to create_database and copy_from Signed-off-by: yihong0618 <zouzou0208@gmail.com> * fix: address comment use opendal Fs Signed-off-by: yihong0618 <zouzou0208@gmail.com> * refactor: make the export mess code clean Signed-off-by: yihong0618 <zouzou0208@gmail.com> --------- Signed-off-by: yihong0618 <zouzou0208@gmail.com>	2025-03-11 08:56:59 +00:00
jeremyhi	1d87bd2d43	feat: alter region follower (#5676 ) * feat: add region follower manager * feat: add region procudure * refactor: make add, remove follower procedure look nice * feat: add region follower procedure * chore: undo some chane, possibly made by AI * feat: on prepare cheking * feat: on update metadata * feat: on broadcast * chore: unit test * feat: add remove follower operation * feat: add or remove region follower procedure * chore: ut * chore: rename * chore: by comment * chore: by comment --------- Co-authored-by: jeremy <jeremy@greptime.local>	2025-03-11 08:44:50 +00:00
jeremyhi	ababeaf538	chore: make memorykv write happily (#5686 ) chore: make memorykv write happly	2025-03-11 07:37:14 +00:00
Lin Yihai	2cbf51d0be	refactor!: Remove `Value::DateTime` and `ValueRef::DateTime`. (#5616 ) * refactor: Remove Value::DateTime and ValueRef::DateTime * fix: don't panic if arrow cast field. * fix: map `ColumnDataType::Datetime` to `ConcreteDataType::timestamp_microsecond_datatype` * fix: Map `ValueData::DatetimeValue` correctly. * refactor: Replace `datetime` with `timestamp_micro_second`	2025-03-11 07:03:27 +00:00
Yingwen	3059b04b19	feat: add a gauge for download tasks (#5681 )	2025-03-11 06:55:13 +00:00
Yingwen	352b197be4	feat: add hint for logical region in RegionScanner (#5684 ) * feat: add a flag to check logical region * feat: sets logical region hint in metric engine * refactor: rename to logical_region	2025-03-11 06:34:39 +00:00