docs: how to get jeprof from target dir

build: add jeprof to linux image
build: add jeprof to image
2025-12-22 22:20:02 +00:00 · 2024-11-14 14:21:48 +08:00 · 2024-11-14 14:21:36 +08:00 · 2024-11-13 23:04:43 +08:00 · 2024-11-13 14:29:44 +00:00 · 2024-11-13 11:15:37 +00:00
223 changed files with 8488 additions and 1780 deletions
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1041,7 +1041,7 @@ dependencies = [
 "bitflags 2.6.0",
 "cexpr",
 "clang-sys",
- "itertools 0.12.1",
+ "itertools 0.10.5",
 "lazy_static",
 "lazycell",
 "log",
@@ -1840,6 +1840,7 @@ dependencies = [
 "common-grpc",
 "common-macro",
 "common-meta",
+ "common-options",
 "common-procedure",
 "common-query",
 "common-recordbatch",
@@ -1934,10 +1935,13 @@ dependencies = [
 "bytes",
 "common-error",
 "common-macro",
+ "common-test-util",
 "futures",
 "paste",
+ "pin-project",
 "serde",
 "snafu 0.8.5",
+ "tokio",
 "toml 0.8.19",
 "zeroize",
 ]
@@ -2070,6 +2074,8 @@ dependencies = [
 "datafusion",
 "datatypes",
 "derive_more",
+ "geo",
+ "geo-types",
 "geohash",
 "h3o",
 "jsonb",
@@ -2088,6 +2094,7 @@ dependencies = [
 "store-api",
 "table",
 "tokio",
+ "wkt",
 ]

 [[package]]
@@ -2236,6 +2243,15 @@ dependencies = [
 "uuid",
 ]

+[[package]]
+name = "common-options"
+version = "0.9.5"
+dependencies = [
+ "common-grpc",
+ "humantime-serde",
+ "serde",
+]
+
 [[package]]
 name = "common-plugins"
 version = "0.9.5"
@@ -3322,6 +3338,8 @@ dependencies = [
 "serde",
 "serde_json",
 "snafu 0.8.5",
+ "sqlparser 0.45.0 (git+https://github.com/GreptimeTeam/sqlparser-rs.git?rev=54a267ac89c09b11c0c88934690530807185d3e7)",
+ "sqlparser_derive 0.1.1",
 ]

 [[package]]
@@ -3706,6 +3724,16 @@ version = "1.0.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "0d6ef0072f8a535281e4876be788938b528e9a1d43900b82c2569af7da799125"

+[[package]]
+name = "earcutr"
+version = "0.4.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "79127ed59a85d7687c409e9978547cffb7dc79675355ed22da6b66fd5f6ead01"
+dependencies = [
+ "itertools 0.11.0",
+ "num-traits",
+]
+
 [[package]]
 name = "either"
 version = "1.13.0"
@@ -4014,6 +4042,12 @@ dependencies = [
 "libc",
 ]

+[[package]]
+name = "float_next_after"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8bf7cc16383c4b8d58b9905a8509f02926ce3058053c056376248d958c9df1e8"
+
 [[package]]
 name = "flow"
 version = "0.9.5"
@@ -4154,6 +4188,7 @@ dependencies = [
 "common-grpc",
 "common-macro",
 "common-meta",
+ "common-options",
 "common-procedure",
 "common-query",
 "common-recordbatch",
@@ -4438,6 +4473,24 @@ dependencies = [
 "version_check",
 ]

+[[package]]
+name = "geo"
+version = "0.29.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "81d088357a9cc60cec8253b3578f6834b4a3aa20edb55f5d1c030c36d8143f11"
+dependencies = [
+ "earcutr",
+ "float_next_after",
+ "geo-types",
+ "geographiclib-rs",
+ "i_overlay",
+ "log",
+ "num-traits",
+ "robust",
+ "rstar",
+ "spade",
+]
+
 [[package]]
 name = "geo-types"
 version = "0.7.13"
@@ -4446,9 +4499,19 @@ checksum = "9ff16065e5720f376fbced200a5ae0f47ace85fd70b7e54269790281353b6d61"
 dependencies = [
 "approx 0.5.1",
 "num-traits",
+ "rstar",
 "serde",
 ]

+[[package]]
+name = "geographiclib-rs"
+version = "0.2.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e6e5ed84f8089c70234b0a8e0aedb6dc733671612ddc0d37c6066052f9781960"
+dependencies = [
+ "libm",
+]
+
 [[package]]
 name = "geohash"
 version = "0.13.1"
@@ -4531,7 +4594,7 @@ dependencies = [
 [[package]]
 name = "greptime-proto"
 version = "0.1.0"
-source = "git+https://github.com/GreptimeTeam/greptime-proto.git?rev=255f87a3318ace3f88a67f76995a0e14910983f4#255f87a3318ace3f88a67f76995a0e14910983f4"
+source = "git+https://github.com/GreptimeTeam/greptime-proto.git?rev=75c5fb569183bb3d0fa1023df9c2214df722b9b1#75c5fb569183bb3d0fa1023df9c2214df722b9b1"
 dependencies = [
 "prost 0.12.6",
 "serde",
@@ -4597,6 +4660,15 @@ dependencies = [
 "num-traits",
 ]

+[[package]]
+name = "hash32"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "47d60b12902ba28e2730cd37e95b8c9223af2808df9e902d4df49588d1470606"
+dependencies = [
+ "byteorder",
+]
+
 [[package]]
 name = "hashbrown"
 version = "0.12.3"
@@ -4692,6 +4764,16 @@ dependencies = [
 "http 1.1.0",
 ]

+[[package]]
+name = "heapless"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0bfb9eb618601c89945a70e254898da93b13be0388091d42117462b265bb3fad"
+dependencies = [
+ "hash32",
+ "stable_deref_trait",
+]
+
 [[package]]
 name = "heck"
 version = "0.4.1"
@@ -5008,7 +5090,7 @@ dependencies = [
 "httpdate",
 "itoa",
 "pin-project-lite",
- "socket2 0.5.7",
+ "socket2 0.4.10",
 "tokio",
 "tower-service",
 "tracing",
@@ -5117,6 +5199,50 @@ dependencies = [
 "tracing",
 ]

+[[package]]
+name = "i_float"
+version = "1.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f5fe043aae28ce70bd2f78b2f5f82a3654d63607c82594da4dabb8b6cb81f2b2"
+dependencies = [
+ "serde",
+]
+
+[[package]]
+name = "i_key_sort"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "347c253b4748a1a28baf94c9ce133b6b166f08573157e05afe718812bc599fcd"
+
+[[package]]
+name = "i_overlay"
+version = "1.7.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a469f68cb8a7cef375b2b0f581faf5859b4b50600438c00d46b71acc25ebbd0c"
+dependencies = [
+ "i_float",
+ "i_key_sort",
+ "i_shape",
+ "i_tree",
+ "rayon",
+]
+
+[[package]]
+name = "i_shape"
+version = "1.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1b44852d57a991c7dedaf76c55bc44f677f547ff899a430d29e13efd6133d7d8"
+dependencies = [
+ "i_float",
+ "serde",
+]
+
+[[package]]
+name = "i_tree"
+version = "0.8.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "155181bc97d770181cf9477da51218a19ee92a8e5be642e796661aee2b601139"
+
 [[package]]
 name = "iana-time-zone"
 version = "0.1.61"
@@ -5523,8 +5649,8 @@ dependencies = [

 [[package]]
 name = "jsonb"
-version = "0.4.1"
-source = "git+https://github.com/databendlabs/jsonb.git?rev=46ad50fc71cf75afbf98eec455f7892a6387c1fc#46ad50fc71cf75afbf98eec455f7892a6387c1fc"
+version = "0.4.3"
+source = "git+https://github.com/databendlabs/jsonb.git?rev=8c8d2fc294a39f3ff08909d60f718639cfba3875#8c8d2fc294a39f3ff08909d60f718639cfba3875"
 dependencies = [
 "byteorder",
 "fast-float",
@@ -5551,6 +5677,19 @@ dependencies = [
 "thiserror",
 ]

+[[package]]
+name = "jsonpath-rust"
+version = "0.7.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "69a61b87f6a55cc6c28fed5739dd36b9642321ce63e4a5e4a4715d69106f4a10"
+dependencies = [
+ "pest",
+ "pest_derive",
+ "regex",
+ "serde_json",
+ "thiserror",
+]
+
 [[package]]
 name = "jsonptr"
 version = "0.4.7"
@@ -5661,7 +5800,7 @@ dependencies = [
 "hyper-rustls",
 "hyper-timeout 0.5.1",
 "hyper-util",
- "jsonpath-rust",
+ "jsonpath-rust 0.5.1",
 "k8s-openapi",
 "kube-core",
 "pem 3.0.4",
@@ -5941,7 +6080,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "4979f22fdb869068da03c9f7528f8297c6fd2606bc3a4affe42e6a823fdb8da4"
 dependencies = [
 "cfg-if",
- "windows-targets 0.52.6",
+ "windows-targets 0.48.5",
 ]

 [[package]]
@@ -6345,6 +6484,7 @@ dependencies = [
 "common-telemetry",
 "datatypes",
 "futures",
+ "futures-util",
 "humantime-serde",
 "meta-srv",
 "rand",
@@ -6375,6 +6515,7 @@ dependencies = [
 "common-grpc",
 "common-macro",
 "common-meta",
+ "common-options",
 "common-procedure",
 "common-procedure-test",
 "common-runtime",
@@ -6556,6 +6697,7 @@ dependencies = [
 "common-error",
 "common-function",
 "common-macro",
+ "common-meta",
 "common-procedure-test",
 "common-query",
 "common-recordbatch",
@@ -7596,6 +7738,7 @@ dependencies = [
 "file-engine",
 "futures",
 "futures-util",
+ "jsonb",
 "lazy_static",
 "meta-client",
 "meter-core",
@@ -8198,6 +8341,7 @@ dependencies = [
 "greptime-proto",
 "itertools 0.10.5",
 "jsonb",
+ "jsonpath-rust 0.7.3",
 "lazy_static",
 "moka",
 "once_cell",
@@ -8396,8 +8540,7 @@ checksum = "439ee305def115ba05938db6eb1644ff94165c5ab5e9420d1c1bcedbba909391"
 [[package]]
 name = "pprof"
 version = "0.13.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ef5c97c51bd34c7e742402e216abdeb44d415fbe6ae41d56b114723e953711cb"
+source = "git+https://github.com/GreptimeTeam/pprof-rs?rev=1bd1e21#1bd1e210d8626da3d1e5aff976e6feee994f576d"
 dependencies = [
 "backtrace",
 "cfg-if",
@@ -8679,7 +8822,7 @@ checksum = "22505a5c94da8e3b7c2996394d1c933236c4d743e81a410bcca4e6989fc066a4"
 dependencies = [
 "bytes",
 "heck 0.5.0",
- "itertools 0.12.1",
+ "itertools 0.10.5",
 "log",
 "multimap",
 "once_cell",
@@ -8731,7 +8874,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "81bddcdb20abf9501610992b6759a4c888aef7d1a7247ef75e2404275ac24af1"
 dependencies = [
 "anyhow",
- "itertools 0.12.1",
+ "itertools 0.10.5",
 "proc-macro2",
 "quote",
 "syn 2.0.79",
@@ -8847,6 +8990,8 @@ dependencies = [
 "auto_impl",
 "base64 0.21.7",
 "bitflags 2.6.0",
+ "bytes",
+ "common-base",
 "common-error",
 "common-macro",
 "common-runtime",
@@ -8891,7 +9036,7 @@ dependencies = [
 "indoc",
 "libc",
 "memoffset 0.9.1",
- "parking_lot 0.12.3",
+ "parking_lot 0.11.2",
 "portable-atomic",
 "pyo3-build-config",
 "pyo3-ffi",
@@ -9568,6 +9713,12 @@ dependencies = [
 "syn 1.0.109",
 ]

+[[package]]
+name = "robust"
+version = "1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cbf4a6aa5f6d6888f39e980649f3ad6b666acdce1d78e95b8a2cb076e687ae30"
+
 [[package]]
 name = "ron"
 version = "0.7.1"
@@ -9662,6 +9813,17 @@ dependencies = [
 "zstd 0.13.2",
 ]

+[[package]]
+name = "rstar"
+version = "0.12.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "133315eb94c7b1e8d0cb097e5a710d850263372fd028fff18969de708afc7008"
+dependencies = [
+ "heapless",
+ "num-traits",
+ "smallvec",
+]
+
 [[package]]
 name = "rstest"
 version = "0.21.0"
@@ -11154,6 +11316,18 @@ dependencies = [
 "windows-sys 0.52.0",
 ]

+[[package]]
+name = "spade"
+version = "2.12.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "93f5ef1f863aca7d1d7dda7ccfc36a0a4279bd6d3c375176e5e0712e25cb4889"
+dependencies = [
+ "hashbrown 0.14.5",
+ "num-traits",
+ "robust",
+ "smallvec",
+]
+
 [[package]]
 name = "sparsevec"
 version = "0.2.0"
@@ -11500,6 +11674,7 @@ dependencies = [
 "derive_builder 0.12.0",
 "futures",
 "humantime",
+ "prost 0.12.6",
 "serde",
 "serde_json",
 "snafu 0.8.5",
@@ -12196,6 +12371,7 @@ dependencies = [
 "frontend",
 "futures",
 "futures-util",
+ "hex",
 "itertools 0.10.5",
 "meta-client",
 "meta-srv",
@@ -13805,7 +13981,7 @@ version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "cf221c93e13a30d793f7645a0e7762c55d169dbb0a49671918a2319d289b10bb"
 dependencies = [
- "windows-sys 0.59.0",
+ "windows-sys 0.48.0",
 ]

 [[package]]
@@ -14147,6 +14323,18 @@ dependencies = [
 "winapi",
 ]

+[[package]]
+name = "wkt"
+version = "0.11.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "54f7f1ff4ea4c18936d6cd26a6fd24f0003af37e951a8e0e8b9e9a2d0bd0a46d"
+dependencies = [
+ "geo-types",
+ "log",
+ "num-traits",
+ "thiserror",
+]
+
 [[package]]
 name = "wyz"
 version = "0.5.1"
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -2,23 +2,25 @@
 members = [
    "src/api",
    "src/auth",
-    "src/catalog",
    "src/cache",
+    "src/catalog",
    "src/client",
    "src/cmd",
    "src/common/base",
    "src/common/catalog",
    "src/common/config",
    "src/common/datasource",
+    "src/common/decimal",
    "src/common/error",
    "src/common/frontend",
    "src/common/function",
-    "src/common/macro",
    "src/common/greptimedb-telemetry",
    "src/common/grpc",
    "src/common/grpc-expr",
+    "src/common/macro",
    "src/common/mem-prof",
    "src/common/meta",
+    "src/common/options",
    "src/common/plugins",
    "src/common/pprof",
    "src/common/procedure",
@@ -30,7 +32,6 @@ members = [
    "src/common/telemetry",
    "src/common/test-util",
    "src/common/time",
-    "src/common/decimal",
    "src/common/version",
    "src/common/wal",
    "src/datanode",
@@ -38,6 +39,7 @@ members = [
    "src/file-engine",
    "src/flow",
    "src/frontend",
+    "src/index",
    "src/log-store",
    "src/meta-client",
    "src/meta-srv",
@@ -57,7 +59,6 @@ members = [
    "src/sql",
    "src/store-api",
    "src/table",
-    "src/index",
    "tests-fuzz",
    "tests-integration",
    "tests/runner",
@@ -117,15 +118,16 @@ datafusion-sql = { git = "https://github.com/waynexia/arrow-datafusion.git", rev
 datafusion-substrait = { git = "https://github.com/waynexia/arrow-datafusion.git", rev = "7823ef2f63663907edab46af0d51359900f608d6" }
 derive_builder = "0.12"
 dotenv = "0.15"
-etcd-client = { version = "0.13" }
+etcd-client = "0.13"
 fst = "0.4.7"
 futures = "0.3"
 futures-util = "0.3"
-greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "255f87a3318ace3f88a67f76995a0e14910983f4" }
+greptime-proto = { git = "https://github.com/GreptimeTeam/greptime-proto.git", rev = "75c5fb569183bb3d0fa1023df9c2214df722b9b1" }
+hex = "0.4"
 humantime = "2.1"
 humantime-serde = "1.1"
 itertools = "0.10"
-jsonb = { git = "https://github.com/databendlabs/jsonb.git", rev = "46ad50fc71cf75afbf98eec455f7892a6387c1fc", default-features = false }
+jsonb = { git = "https://github.com/databendlabs/jsonb.git", rev = "8c8d2fc294a39f3ff08909d60f718639cfba3875", default-features = false }
 lazy_static = "1.4"
 meter-core = { git = "https://github.com/GreptimeTeam/greptime-meter.git", rev = "a10facb353b41460eeb98578868ebf19c2084fac" }
 mockall = "0.11.4"
@@ -151,7 +153,7 @@ raft-engine = { version = "0.4.1", default-features = false }
 rand = "0.8"
 ratelimit = "0.9"
 regex = "1.8"
-regex-automata = { version = "0.4" }
+regex-automata = "0.4"
 reqwest = { version = "0.12", default-features = false, features = [
    "json",
    "rustls-tls-native-roots",
@@ -182,11 +184,11 @@ strum = { version = "0.25", features = ["derive"] }
 tempfile = "3"
 tokio = { version = "1.40", features = ["full"] }
 tokio-postgres = "0.7"
-tokio-stream = { version = "0.1" }
+tokio-stream = "0.1"
 tokio-util = { version = "0.7", features = ["io-util", "compat"] }
 toml = "0.8.8"
 tonic = { version = "0.11", features = ["tls", "gzip", "zstd"] }
-tower = { version = "0.4" }
+tower = "0.4"
 tracing-appender = "0.2"
 tracing-subscriber = { version = "0.3", features = ["env-filter", "json", "fmt"] }
 typetag = "0.2"
@@ -214,6 +216,7 @@ common-grpc-expr = { path = "src/common/grpc-expr" }
 common-macro = { path = "src/common/macro" }
 common-mem-prof = { path = "src/common/mem-prof" }
 common-meta = { path = "src/common/meta" }
+common-options = { path = "src/common/options" }
 common-plugins = { path = "src/common/plugins" }
 common-pprof = { path = "src/common/pprof" }
 common-procedure = { path = "src/common/procedure" }
@@ -261,6 +264,8 @@ tokio-rustls = { git = "https://github.com/GreptimeTeam/tokio-rustls" }
 # This is commented, since we are not using aws-lc-sys, if we need to use it, we need to uncomment this line or use a release after this commit, or it wouldn't compile with gcc < 8.1
 # see https://github.com/aws/aws-lc-rs/pull/526
 # aws-lc-sys = { git ="https://github.com/aws/aws-lc-rs", rev = "556558441e3494af4b156ae95ebc07ebc2fd38aa" }
+# Apply a fix for pprof for unaligned pointer access
+pprof = { git = "https://github.com/GreptimeTeam/pprof-rs", rev = "1bd1e21" }

 [workspace.dependencies.meter-macros]
 git = "https://github.com/GreptimeTeam/greptime-meter.git"
--- a/README.md
+++ b/README.md
@@ -6,7 +6,7 @@
  </picture>
 </p>

-<h2 align="center">Unified Time Series Database for Metrics, Logs, and Events</h2>
+<h2 align="center">Unified & Cost-Effective Time Series Database for Metrics, Logs, and Events</h2>

 <div align="center">
 <h3 align="center">
@@ -48,9 +48,21 @@
 </a>
 </div>

+- [Introduction](#introduction)
+- [**Features: Why GreptimeDB**](#why-greptimedb)
+- [Architecture](https://docs.greptime.com/contributor-guide/overview/#architecture)
+- [Try it for free](#try-greptimedb)
+- [Getting Started](#getting-started)
+- [Project Status](#project-status)
+- [Join the community](#community)
+  - [Contributing](#contributing)
+- [Extension](#extension )
+- [License](#license)
+- [Acknowledgement](#acknowledgement)
+
 ## Introduction

-**GreptimeDB** is an open-source unified time-series database for **Metrics**, **Logs**, and **Events** (also **Traces** in plan). You can gain real-time insights from Edge to Cloud at any scale.
+**GreptimeDB** is an open-source unified & cost-effective time-series database for **Metrics**, **Logs**, and **Events** (also **Traces** in plan). You can gain real-time insights from Edge to Cloud at Any Scale.

 ## Why GreptimeDB

--- a/config/config.md
+++ b/config/config.md
@@ -93,8 +93,8 @@
 | `storage` | -- | -- | The data storage options. |
 | `storage.data_home` | String | `/tmp/greptimedb/` | 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 | Cache configuration for object storage such as 'S3' etc.<br/>The local file cache directory. |
-| `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. |
+| `storage.cache_path` | String | Unset | Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.<br/>The local file cache directory. |
+| `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. |
 | `storage.bucket` | String | Unset | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
 | `storage.root` | String | Unset | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
 | `storage.access_key_id` | String | Unset | The access key id of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3` and `Oss`**. |
@@ -126,9 +126,9 @@
 | `region_engine.mito.vector_cache_size` | String | Auto | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
 | `region_engine.mito.page_cache_size` | String | Auto | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
 | `region_engine.mito.selector_result_cache_size` | String | Auto | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
-| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. |
+| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance. |
 | `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}/write_cache`. |
-| `region_engine.mito.experimental_write_cache_size` | String | `512MB` | Capacity for write cache. |
+| `region_engine.mito.experimental_write_cache_size` | String | `1GiB` | Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger. |
 | `region_engine.mito.experimental_write_cache_ttl` | String | Unset | TTL for write cache. |
 | `region_engine.mito.sst_write_buffer_size` | String | `8MB` | Buffer size for SST writing. |
 | `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
@@ -416,8 +416,8 @@
 | `storage` | -- | -- | The data storage options. |
 | `storage.data_home` | String | `/tmp/greptimedb/` | 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 | Cache configuration for object storage such as 'S3' etc.<br/>The local file cache directory. |
-| `storage.cache_capacity` | String | Unset | The local file cache capacity in bytes. |
+| `storage.cache_path` | String | Unset | Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.<br/>The local file cache directory. |
+| `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. |
 | `storage.bucket` | String | Unset | The S3 bucket name.<br/>**It's only used when the storage type is `S3`, `Oss` and `Gcs`**. |
 | `storage.root` | String | Unset | The S3 data will be stored in the specified prefix, for example, `s3://${bucket}/${root}`.<br/>**It's only used when the storage type is `S3`, `Oss` and `Azblob`**. |
 | `storage.access_key_id` | String | Unset | The access key id of the aws account.<br/>It's **highly recommended** to use AWS IAM roles instead of hardcoding the access key id and secret key.<br/>**It's only used when the storage type is `S3` and `Oss`**. |
@@ -449,9 +449,9 @@
 | `region_engine.mito.vector_cache_size` | String | Auto | Cache size for vectors and arrow arrays. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
 | `region_engine.mito.page_cache_size` | String | Auto | Cache size for pages of SST row groups. Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/8 of OS memory. |
 | `region_engine.mito.selector_result_cache_size` | String | Auto | Cache size for time series selector (e.g. `last_value()`). Setting it to 0 to disable the cache.<br/>If not set, it's default to 1/16 of OS memory with a max limitation of 512MB. |
-| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. |
+| `region_engine.mito.enable_experimental_write_cache` | Bool | `false` | Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance. |
 | `region_engine.mito.experimental_write_cache_path` | String | `""` | File system path for write cache, defaults to `{data_home}/write_cache`. |
-| `region_engine.mito.experimental_write_cache_size` | String | `512MB` | Capacity for write cache. |
+| `region_engine.mito.experimental_write_cache_size` | String | `1GiB` | Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger. |
 | `region_engine.mito.experimental_write_cache_ttl` | String | Unset | TTL for write cache. |
 | `region_engine.mito.sst_write_buffer_size` | String | `8MB` | Buffer size for SST writing. |
 | `region_engine.mito.scan_parallelism` | Integer | `0` | Parallelism to scan a region (default: 1/4 of cpu cores).<br/>- `0`: using the default value (1/4 of cpu cores).<br/>- `1`: scan in current thread.<br/>- `n`: scan in parallelism n. |
--- a/config/datanode.example.toml
+++ b/config/datanode.example.toml
@@ -294,14 +294,14 @@ data_home = "/tmp/greptimedb/"
 ## - `Oss`: the data is stored in the Aliyun OSS.
 type = "File"

-## Cache configuration for object storage such as 'S3' etc.
+## Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.
 ## The local file cache directory.
 ## @toml2docs:none-default
 cache_path = "/path/local_cache"

-## The local file cache capacity in bytes.
+## The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger.
 ## @toml2docs:none-default
-cache_capacity = "256MB"
+cache_capacity = "1GiB"

 ## The S3 bucket name.
 ## **It's only used when the storage type is `S3`, `Oss` and `Gcs`**.
@@ -459,14 +459,14 @@ auto_flush_interval = "1h"
 ## @toml2docs:none-default="Auto"
 #+ selector_result_cache_size = "512MB"

-## Whether to enable the experimental write cache.
+## Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance.
 enable_experimental_write_cache = false

 ## File system path for write cache, defaults to `{data_home}/write_cache`.
 experimental_write_cache_path = ""

-## Capacity for write cache.
-experimental_write_cache_size = "512MB"
+## Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger.
+experimental_write_cache_size = "1GiB"

 ## TTL for write cache.
 ## @toml2docs:none-default
--- a/config/standalone.example.toml
+++ b/config/standalone.example.toml
@@ -332,14 +332,14 @@ data_home = "/tmp/greptimedb/"
 ## - `Oss`: the data is stored in the Aliyun OSS.
 type = "File"

-## Cache configuration for object storage such as 'S3' etc.
+## Cache configuration for object storage such as 'S3' etc. It is recommended to configure it when using object storage for better performance.
 ## The local file cache directory.
 ## @toml2docs:none-default
 cache_path = "/path/local_cache"

-## The local file cache capacity in bytes.
+## The local file cache capacity in bytes. If your disk space is sufficient, it is recommended to set it larger.
 ## @toml2docs:none-default
-cache_capacity = "256MB"
+cache_capacity = "1GiB"

 ## The S3 bucket name.
 ## **It's only used when the storage type is `S3`, `Oss` and `Gcs`**.
@@ -497,14 +497,14 @@ auto_flush_interval = "1h"
 ## @toml2docs:none-default="Auto"
 #+ selector_result_cache_size = "512MB"

-## Whether to enable the experimental write cache.
+## Whether to enable the experimental write cache. It is recommended to enable it when using object storage for better performance.
 enable_experimental_write_cache = false

 ## File system path for write cache, defaults to `{data_home}/write_cache`.
 experimental_write_cache_path = ""

-## Capacity for write cache.
-experimental_write_cache_size = "512MB"
+## Capacity for write cache. If your disk space is sufficient, it is recommended to set it larger.
+experimental_write_cache_size = "1GiB"

 ## TTL for write cache.
 ## @toml2docs:none-default
--- a/docker/buildx/centos/Dockerfile
+++ b/docker/buildx/centos/Dockerfile
@@ -49,6 +49,7 @@ RUN yum install -y epel-release \

 WORKDIR /greptime
 COPY --from=builder /out/target/${OUTPUT_DIR}/greptime /greptime/bin/
+COPY --from=builder /out/target/${OUTPUT_DIR}/build/tikv-jemalloc-sys-*/out/build/bin/jeprof /greptime/bin/
 ENV PATH /greptime/bin/:$PATH

 ENTRYPOINT ["greptime"]
--- a/docker/buildx/ubuntu/Dockerfile
+++ b/docker/buildx/ubuntu/Dockerfile
@@ -57,6 +57,9 @@ RUN python3 -m pip install -r /etc/greptime/requirements.txt

 WORKDIR /greptime
 COPY --from=builder /out/target/${OUTPUT_DIR}/greptime /greptime/bin/
+COPY --from=builder /out/target/${OUTPUT_DIR}/build/tikv-jemalloc-sys-*/out/build/bin/jeprof /greptime/bin/
+RUN chmod +x /greptime/bin/jeprof
+
 ENV PATH /greptime/bin/:$PATH

 ENTRYPOINT ["greptime"]
--- a/docs/how-to/how-to-change-log-level-on-the-fly.md
+++ b/docs/how-to/how-to-change-log-level-on-the-fly.md
@@ -4,13 +4,13 @@

 example:
 ```bash
-curl --data "trace;flow=debug" 127.0.0.1:4000/debug/log_level
+curl --data "trace,flow=debug" 127.0.0.1:4000/debug/log_level
 ```
 And database will reply with something like:
 ```bash
-Log Level changed from Some("info") to "trace;flow=debug"%
+Log Level changed from Some("info") to "trace,flow=debug"%
 ```

-The data is a string in the format of `global_level;module1=level1;module2=level2;...` that follow the same rule of `RUST_LOG`. 
+The data is a string in the format of `global_level,module1=level1,module2=level2,...` that follow the same rule of `RUST_LOG`. 

 The module is the module name of the log, and the level is the log level. The log level can be one of the following: `trace`, `debug`, `info`, `warn`, `error`, `off`(case insensitive).
--- a/docs/how-to/how-to-profile-memory.md
+++ b/docs/how-to/how-to-profile-memory.md
@@ -4,6 +4,18 @@ This crate provides an easy approach to dump memory profiling info.

 ## Prerequisites
 ### jemalloc
+jeprof is already compiled in the target directory of GreptimeDB. You can find the binary and use it.
+```
+# find jeprof binary
+find . -name 'jeprof'
+
+# add executable permission
+chmod +x <path_to_jeprof>
+```
+The path is usually under `./target/${PROFILE}/build/tikv-jemalloc-sys-${HASH}/out/build/bin/jeprof`.
+
+The default version of jemalloc installed from the package manager may not have the `--collapsed` option.
+You may need to check the whether the `jeprof` version is >= `5.3.0` if you want to install it from the package manager.
 ```bash
 # for macOS
 brew install jemalloc
--- a/src/api/src/helper.rs
+++ b/src/api/src/helper.rs
@@ -36,15 +36,14 @@ use datatypes::vectors::{
    TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, UInt32Vector,
    UInt64Vector, VectorRef,
 };
-use greptime_proto::v1;
 use greptime_proto::v1::column_data_type_extension::TypeExt;
 use greptime_proto::v1::ddl_request::Expr;
 use greptime_proto::v1::greptime_request::Request;
 use greptime_proto::v1::query_request::Query;
 use greptime_proto::v1::value::ValueData;
 use greptime_proto::v1::{
-    ColumnDataTypeExtension, DdlRequest, DecimalTypeExtension, JsonTypeExtension, QueryRequest,
-    Row, SemanticType,
+    self, ColumnDataTypeExtension, DdlRequest, DecimalTypeExtension, JsonTypeExtension,
+    QueryRequest, Row, SemanticType, VectorTypeExtension,
 };
 use paste::paste;
 use snafu::prelude::*;
@@ -150,6 +149,17 @@ impl From<ColumnDataTypeWrapper> for ConcreteDataType {
                    ConcreteDataType::decimal128_default_datatype()
                }
            }
+            ColumnDataType::Vector => {
+                if let Some(TypeExt::VectorType(d)) = datatype_wrapper
+                    .datatype_ext
+                    .as_ref()
+                    .and_then(|datatype_ext| datatype_ext.type_ext.as_ref())
+                {
+                    ConcreteDataType::vector_datatype(d.dim)
+                } else {
+                    ConcreteDataType::vector_default_datatype()
+                }
+            }
        }
    }
 }
@@ -231,6 +241,15 @@ impl ColumnDataTypeWrapper {
            }),
        }
    }
+
+    pub fn vector_datatype(dim: u32) -> Self {
+        ColumnDataTypeWrapper {
+            datatype: ColumnDataType::Vector,
+            datatype_ext: Some(ColumnDataTypeExtension {
+                type_ext: Some(TypeExt::VectorType(VectorTypeExtension { dim })),
+            }),
+        }
+    }
 }

 impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
@@ -249,7 +268,7 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
            ConcreteDataType::UInt64(_) => ColumnDataType::Uint64,
            ConcreteDataType::Float32(_) => ColumnDataType::Float32,
            ConcreteDataType::Float64(_) => ColumnDataType::Float64,
-            ConcreteDataType::Binary(_) | ConcreteDataType::Json(_) => ColumnDataType::Binary,
+            ConcreteDataType::Binary(_) => ColumnDataType::Binary,
            ConcreteDataType::String(_) => ColumnDataType::String,
            ConcreteDataType::Date(_) => ColumnDataType::Date,
            ConcreteDataType::DateTime(_) => ColumnDataType::Datetime,
@@ -271,6 +290,8 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
                IntervalType::MonthDayNano(_) => ColumnDataType::IntervalMonthDayNano,
            },
            ConcreteDataType::Decimal128(_) => ColumnDataType::Decimal128,
+            ConcreteDataType::Json(_) => ColumnDataType::Json,
+            ConcreteDataType::Vector(_) => ColumnDataType::Vector,
            ConcreteDataType::Null(_)
            | ConcreteDataType::List(_)
            | ConcreteDataType::Dictionary(_)
@@ -289,15 +310,17 @@ impl TryFrom<ConcreteDataType> for ColumnDataTypeWrapper {
                        })),
                    })
            }
-            ColumnDataType::Binary => {
-                if datatype == ConcreteDataType::json_datatype() {
-                    // Json is the same as  binary in proto. The extension marks the binary in proto is actually a json.
-                    Some(ColumnDataTypeExtension {
-                        type_ext: Some(TypeExt::JsonType(JsonTypeExtension::JsonBinary.into())),
+            ColumnDataType::Json => datatype.as_json().map(|_| ColumnDataTypeExtension {
+                type_ext: Some(TypeExt::JsonType(JsonTypeExtension::JsonBinary.into())),
+            }),
+            ColumnDataType::Vector => {
+                datatype
+                    .as_vector()
+                    .map(|vector_type| ColumnDataTypeExtension {
+                        type_ext: Some(TypeExt::VectorType(VectorTypeExtension {
+                            dim: vector_type.dim as _,
+                        })),
                    })
-                } else {
-                    None
-                }
            }
            _ => None,
        };
@@ -422,6 +445,10 @@ pub fn values_with_capacity(datatype: ColumnDataType, capacity: usize) -> Values
            string_values: Vec::with_capacity(capacity),
            ..Default::default()
        },
+        ColumnDataType::Vector => Values {
+            binary_values: Vec::with_capacity(capacity),
+            ..Default::default()
+        },
    }
 }

@@ -673,6 +700,7 @@ pub fn pb_values_to_vector_ref(data_type: &ConcreteDataType, values: Values) ->
                Decimal128::from_value_precision_scale(x.hi, x.lo, d.precision(), d.scale()).into()
            }),
        )),
+        ConcreteDataType::Vector(_) => Arc::new(BinaryVector::from_vec(values.binary_values)),
        ConcreteDataType::Null(_)
        | ConcreteDataType::List(_)
        | ConcreteDataType::Dictionary(_)
@@ -838,6 +866,7 @@ pub fn pb_values_to_values(data_type: &ConcreteDataType, values: Values) -> Vec<
                ))
            })
            .collect(),
+        ConcreteDataType::Vector(_) => values.binary_values.into_iter().map(|v| v.into()).collect(),
        ConcreteDataType::Null(_)
        | ConcreteDataType::List(_)
        | ConcreteDataType::Dictionary(_)
@@ -862,10 +891,7 @@ pub fn is_column_type_value_eq(
    ColumnDataTypeWrapper::try_new(type_value, type_extension)
        .map(|wrapper| {
            let datatype = ConcreteDataType::from(wrapper);
-            (datatype == *expect_type)
-            // Json type leverage binary type in pb, so this is valid.
-                || (datatype == ConcreteDataType::binary_datatype()
-                    && *expect_type == ConcreteDataType::json_datatype())
+            expect_type == &datatype
        })
        .unwrap_or(false)
 }
@@ -1152,6 +1178,10 @@ mod tests {
        let values = values_with_capacity(ColumnDataType::Decimal128, 2);
        let values = values.decimal128_values;
        assert_eq!(2, values.capacity());
+
+        let values = values_with_capacity(ColumnDataType::Vector, 2);
+        let values = values.binary_values;
+        assert_eq!(2, values.capacity());
    }

    #[test]
@@ -1239,7 +1269,11 @@ mod tests {
        assert_eq!(
            ConcreteDataType::decimal128_datatype(10, 2),
            ColumnDataTypeWrapper::decimal128_datatype(10, 2).into()
-        )
+        );
+        assert_eq!(
+            ConcreteDataType::vector_datatype(3),
+            ColumnDataTypeWrapper::vector_datatype(3).into()
+        );
    }

    #[test]
@@ -1335,6 +1369,10 @@ mod tests {
                .try_into()
                .unwrap()
        );
+        assert_eq!(
+            ColumnDataTypeWrapper::vector_datatype(3),
+            ConcreteDataType::vector_datatype(3).try_into().unwrap()
+        );

        let result: Result<ColumnDataTypeWrapper> = ConcreteDataType::null_datatype().try_into();
        assert!(result.is_err());
--- a/src/api/src/v1/column_def.rs
+++ b/src/api/src/v1/column_def.rs
@@ -15,8 +15,10 @@
 use std::collections::HashMap;

 use datatypes::schema::{
-    ColumnDefaultConstraint, ColumnSchema, FulltextOptions, COMMENT_KEY, FULLTEXT_KEY,
+    ColumnDefaultConstraint, ColumnSchema, FulltextAnalyzer, FulltextOptions, COMMENT_KEY,
+    FULLTEXT_KEY, INVERTED_INDEX_KEY,
 };
+use greptime_proto::v1::Analyzer;
 use snafu::ResultExt;

 use crate::error::{self, Result};
@@ -25,6 +27,8 @@ use crate::v1::{ColumnDef, ColumnOptions, SemanticType};

 /// Key used to store fulltext options in gRPC column options.
 const FULLTEXT_GRPC_KEY: &str = "fulltext";
+/// Key used to store inverted index options in gRPC column options.
+const INVERTED_INDEX_GRPC_KEY: &str = "inverted_index";

 /// Tries to construct a `ColumnSchema` from the given  `ColumnDef`.
 pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
@@ -49,10 +53,13 @@ pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
    if !column_def.comment.is_empty() {
        metadata.insert(COMMENT_KEY.to_string(), column_def.comment.clone());
    }
-    if let Some(options) = column_def.options.as_ref()
-        && let Some(fulltext) = options.options.get(FULLTEXT_GRPC_KEY)
-    {
-        metadata.insert(FULLTEXT_KEY.to_string(), fulltext.to_string());
+    if let Some(options) = column_def.options.as_ref() {
+        if let Some(fulltext) = options.options.get(FULLTEXT_GRPC_KEY) {
+            metadata.insert(FULLTEXT_KEY.to_string(), fulltext.clone());
+        }
+        if let Some(inverted_index) = options.options.get(INVERTED_INDEX_GRPC_KEY) {
+            metadata.insert(INVERTED_INDEX_KEY.to_string(), inverted_index.clone());
+        }
    }

    ColumnSchema::new(&column_def.name, data_type.into(), column_def.is_nullable)
@@ -70,7 +77,12 @@ pub fn options_from_column_schema(column_schema: &ColumnSchema) -> Option<Column
    if let Some(fulltext) = column_schema.metadata().get(FULLTEXT_KEY) {
        options
            .options
-            .insert(FULLTEXT_GRPC_KEY.to_string(), fulltext.to_string());
+            .insert(FULLTEXT_GRPC_KEY.to_string(), fulltext.clone());
+    }
+    if let Some(inverted_index) = column_schema.metadata().get(INVERTED_INDEX_KEY) {
+        options
+            .options
+            .insert(INVERTED_INDEX_GRPC_KEY.to_string(), inverted_index.clone());
    }

    (!options.options.is_empty()).then_some(options)
@@ -93,6 +105,14 @@ pub fn options_from_fulltext(fulltext: &FulltextOptions) -> Result<Option<Column
    Ok((!options.options.is_empty()).then_some(options))
 }

+/// Tries to construct a `FulltextAnalyzer` from the given analyzer.
+pub fn as_fulltext_option(analyzer: Analyzer) -> FulltextAnalyzer {
+    match analyzer {
+        Analyzer::English => FulltextAnalyzer::English,
+        Analyzer::Chinese => FulltextAnalyzer::Chinese,
+    }
+}
+
 #[cfg(test)]
 mod tests {

@@ -115,10 +135,13 @@ mod tests {
            comment: "test_comment".to_string(),
            datatype_extension: None,
            options: Some(ColumnOptions {
-                options: HashMap::from([(
-                    FULLTEXT_GRPC_KEY.to_string(),
-                    "{\"enable\":true}".to_string(),
-                )]),
+                options: HashMap::from([
+                    (
+                        FULLTEXT_GRPC_KEY.to_string(),
+                        "{\"enable\":true}".to_string(),
+                    ),
+                    (INVERTED_INDEX_GRPC_KEY.to_string(), "true".to_string()),
+                ]),
            }),
        };

@@ -139,6 +162,7 @@ mod tests {
                ..Default::default()
            }
        );
+        assert!(schema.is_inverted_indexed());
    }

    #[test]
@@ -153,12 +177,17 @@ mod tests {
                analyzer: FulltextAnalyzer::English,
                case_sensitive: false,
            })
-            .unwrap();
+            .unwrap()
+            .set_inverted_index(true);
        let options = options_from_column_schema(&schema).unwrap();
        assert_eq!(
            options.options.get(FULLTEXT_GRPC_KEY).unwrap(),
            "{\"enable\":true,\"analyzer\":\"English\",\"case-sensitive\":false}"
        );
+        assert_eq!(
+            options.options.get(INVERTED_INDEX_GRPC_KEY).unwrap(),
+            "true"
+        );
    }

    #[test]
--- a/src/catalog/src/error.rs
+++ b/src/catalog/src/error.rs
@@ -178,6 +178,12 @@ pub enum Error {
        location: Location,
    },

+    #[snafu(display("Partition manager not found, it's not expected."))]
+    PartitionManagerNotFound {
+        #[snafu(implicit)]
+        location: Location,
+    },
+
    #[snafu(display("Failed to find table partitions"))]
    FindPartitions { source: partition::error::Error },

@@ -301,6 +307,7 @@ impl ErrorExt for Error {
            | Error::CastManager { .. }
            | Error::Json { .. }
            | Error::GetInformationExtension { .. }
+            | Error::PartitionManagerNotFound { .. }
            | Error::ProcedureIdNotFound { .. } => StatusCode::Unexpected,

            Error::ViewPlanColumnsChanged { .. } => StatusCode::InvalidArguments,
--- a/src/catalog/src/system_schema/information_schema/partitions.rs
+++ b/src/catalog/src/system_schema/information_schema/partitions.rs
@@ -34,15 +34,14 @@ use datatypes::vectors::{
 };
 use futures::{StreamExt, TryStreamExt};
 use partition::manager::PartitionInfo;
-use partition::partition::PartitionDef;
 use snafu::{OptionExt, ResultExt};
-use store_api::storage::{RegionId, ScanRequest, TableId};
+use store_api::storage::{ScanRequest, TableId};
 use table::metadata::{TableInfo, TableType};

 use super::PARTITIONS;
 use crate::error::{
-    CreateRecordBatchSnafu, FindPartitionsSnafu, InternalSnafu, Result,
-    UpgradeWeakCatalogManagerRefSnafu,
+    CreateRecordBatchSnafu, FindPartitionsSnafu, InternalSnafu, PartitionManagerNotFoundSnafu,
+    Result, UpgradeWeakCatalogManagerRefSnafu,
 };
 use crate::kvbackend::KvBackendCatalogManager;
 use crate::system_schema::information_schema::{InformationTable, Predicates};
@@ -236,7 +235,8 @@ impl InformationSchemaPartitionsBuilder {
        let partition_manager = catalog_manager
            .as_any()
            .downcast_ref::<KvBackendCatalogManager>()
-            .map(|catalog_manager| catalog_manager.partition_manager());
+            .map(|catalog_manager| catalog_manager.partition_manager())
+            .context(PartitionManagerNotFoundSnafu)?;

        let predicates = Predicates::from_scan_request(&request);

@@ -262,27 +262,10 @@ impl InformationSchemaPartitionsBuilder {
                let table_ids: Vec<TableId> =
                    table_infos.iter().map(|info| info.ident.table_id).collect();

-                let mut table_partitions = if let Some(partition_manager) = &partition_manager {
-                    partition_manager
-                        .batch_find_table_partitions(&table_ids)
-                        .await
-                        .context(FindPartitionsSnafu)?
-                } else {
-                    // Current node must be a standalone instance, contains only one partition by default.
-                    // TODO(dennis): change it when we support multi-regions for standalone.
-                    table_ids
-                        .into_iter()
-                        .map(|table_id| {
-                            (
-                                table_id,
-                                vec![PartitionInfo {
-                                    id: RegionId::new(table_id, 0),
-                                    partition: PartitionDef::new(vec![], vec![]),
-                                }],
-                            )
-                        })
-                        .collect()
-                };
+                let mut table_partitions = partition_manager
+                    .batch_find_table_partitions(&table_ids)
+                    .await
+                    .context(FindPartitionsSnafu)?;

                for table_info in table_infos {
                    let partitions = table_partitions
--- a/src/catalog/src/system_schema/information_schema/tables.rs
+++ b/src/catalog/src/system_schema/information_schema/tables.rs
@@ -12,13 +12,16 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

+use std::collections::HashSet;
 use std::sync::{Arc, Weak};

 use arrow_schema::SchemaRef as ArrowSchemaRef;
-use common_catalog::consts::INFORMATION_SCHEMA_TABLES_TABLE_ID;
+use common_catalog::consts::{INFORMATION_SCHEMA_TABLES_TABLE_ID, MITO_ENGINE};
 use common_error::ext::BoxedError;
+use common_meta::datanode::RegionStat;
 use common_recordbatch::adapter::RecordBatchStreamAdapter;
 use common_recordbatch::{RecordBatch, SendableRecordBatchStream};
+use common_telemetry::error;
 use datafusion::execution::TaskContext;
 use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
 use datafusion::physical_plan::streaming::PartitionStream as DfPartitionStream;
@@ -31,7 +34,7 @@ use datatypes::vectors::{
 };
 use futures::TryStreamExt;
 use snafu::{OptionExt, ResultExt};
-use store_api::storage::{ScanRequest, TableId};
+use store_api::storage::{RegionId, ScanRequest, TableId};
 use table::metadata::{TableInfo, TableType};

 use super::TABLES;
@@ -39,6 +42,7 @@ use crate::error::{
    CreateRecordBatchSnafu, InternalSnafu, Result, UpgradeWeakCatalogManagerRefSnafu,
 };
 use crate::system_schema::information_schema::{InformationTable, Predicates};
+use crate::system_schema::utils;
 use crate::CatalogManager;

 pub const TABLE_CATALOG: &str = "table_catalog";
@@ -234,17 +238,51 @@ impl InformationSchemaTablesBuilder {
            .context(UpgradeWeakCatalogManagerRefSnafu)?;
        let predicates = Predicates::from_scan_request(&request);

+        let information_extension = utils::information_extension(&self.catalog_manager)?;
+
+        // TODO(dennis): `region_stats` API is not stable in distributed cluster because of network issue etc.
+        // But we don't want the statements such as `show tables` fail,
+        // so using `unwrap_or_else` here instead of `?` operator.
+        let region_stats = information_extension
+            .region_stats()
+            .await
+            .map_err(|e| {
+                error!(e; "Failed to call region_stats");
+                e
+            })
+            .unwrap_or_else(|_| vec![]);
+
        for schema_name in catalog_manager.schema_names(&catalog_name, None).await? {
            let mut stream = catalog_manager.tables(&catalog_name, &schema_name, None);

            while let Some(table) = stream.try_next().await? {
                let table_info = table.table_info();
+
+                // TODO(dennis): make it working for metric engine
+                let table_region_stats =
+                    if table_info.meta.engine == MITO_ENGINE || table_info.is_physical_table() {
+                        let region_ids = table_info
+                            .meta
+                            .region_numbers
+                            .iter()
+                            .map(|n| RegionId::new(table_info.ident.table_id, *n))
+                            .collect::<HashSet<_>>();
+
+                        region_stats
+                            .iter()
+                            .filter(|stat| region_ids.contains(&stat.id))
+                            .collect::<Vec<_>>()
+                    } else {
+                        vec![]
+                    };
+
                self.add_table(
                    &predicates,
                    &catalog_name,
                    &schema_name,
                    table_info,
                    table.table_type(),
+                    &table_region_stats,
                );
            }
        }
@@ -260,6 +298,7 @@ impl InformationSchemaTablesBuilder {
        schema_name: &str,
        table_info: Arc<TableInfo>,
        table_type: TableType,
+        region_stats: &[&RegionStat],
    ) {
        let table_name = table_info.name.as_ref();
        let table_id = table_info.table_id();
@@ -273,7 +312,9 @@ impl InformationSchemaTablesBuilder {

        let row = [
            (TABLE_CATALOG, &Value::from(catalog_name)),
+            (TABLE_ID, &Value::from(table_id)),
            (TABLE_SCHEMA, &Value::from(schema_name)),
+            (ENGINE, &Value::from(engine)),
            (TABLE_NAME, &Value::from(table_name)),
            (TABLE_TYPE, &Value::from(table_type_text)),
        ];
@@ -287,21 +328,39 @@ impl InformationSchemaTablesBuilder {
        self.table_names.push(Some(table_name));
        self.table_types.push(Some(table_type_text));
        self.table_ids.push(Some(table_id));
+
+        let data_length = region_stats.iter().map(|stat| stat.sst_size).sum();
+        let table_rows = region_stats.iter().map(|stat| stat.num_rows).sum();
+        let index_length = region_stats.iter().map(|stat| stat.index_size).sum();
+
+        // It's not precise, but it is acceptable for long-term data storage.
+        let avg_row_length = if table_rows > 0 {
+            let total_data_length = data_length
+                + region_stats
+                    .iter()
+                    .map(|stat| stat.memtable_size)
+                    .sum::<u64>();
+
+            total_data_length / table_rows
+        } else {
+            0
+        };
+
+        self.data_length.push(Some(data_length));
+        self.index_length.push(Some(index_length));
+        self.table_rows.push(Some(table_rows));
+        self.avg_row_length.push(Some(avg_row_length));
+
        // TODO(sunng87): use real data for these fields
-        self.data_length.push(Some(0));
        self.max_data_length.push(Some(0));
-        self.index_length.push(Some(0));
-        self.avg_row_length.push(Some(0));
-        self.max_index_length.push(Some(0));
        self.checksum.push(Some(0));
-        self.table_rows.push(Some(0));
+        self.max_index_length.push(Some(0));
        self.data_free.push(Some(0));
        self.auto_increment.push(Some(0));
        self.row_format.push(Some("Fixed"));
        self.table_collation.push(Some("utf8_bin"));
        self.update_time.push(None);
        self.check_time.push(None);
-
        // use mariadb default table version number here
        self.version.push(Some(11));
        self.table_comment.push(table_info.desc.as_deref());
--- a/src/cmd/Cargo.toml
+++ b/src/cmd/Cargo.toml
@@ -33,6 +33,7 @@ common-error.workspace = true
 common-grpc.workspace = true
 common-macro.workspace = true
 common-meta.workspace = true
+common-options.workspace = true
 common-procedure.workspace = true
 common-query.workspace = true
 common-recordbatch.workspace = true
--- a/src/cmd/tests/load_config_test.rs
+++ b/src/cmd/tests/load_config_test.rs
@@ -20,13 +20,13 @@ use common_config::Configurable;
 use common_grpc::channel_manager::{
    DEFAULT_MAX_GRPC_RECV_MESSAGE_SIZE, DEFAULT_MAX_GRPC_SEND_MESSAGE_SIZE,
 };
+use common_options::datanode::{ClientOptions, DatanodeClientOptions};
 use common_telemetry::logging::{LoggingOptions, SlowQueryOptions, DEFAULT_OTLP_ENDPOINT};
 use common_wal::config::raft_engine::RaftEngineConfig;
 use common_wal::config::DatanodeWalConfig;
 use datanode::config::{DatanodeOptions, RegionEngineConfig, StorageConfig};
 use file_engine::config::EngineConfig;
 use frontend::frontend::FrontendOptions;
-use frontend::service_config::datanode::DatanodeClientOptions;
 use meta_client::MetaClientOptions;
 use meta_srv::metasrv::MetasrvOptions;
 use meta_srv::selector::SelectorType;
@@ -126,10 +126,11 @@ fn test_load_frontend_example_config() {
                tracing_sample_ratio: Some(Default::default()),
                ..Default::default()
            },
-            datanode: frontend::service_config::DatanodeOptions {
-                client: DatanodeClientOptions {
+            datanode: DatanodeClientOptions {
+                client: ClientOptions {
                    connect_timeout: Duration::from_secs(10),
                    tcp_nodelay: true,
+                    ..Default::default()
                },
            },
            export_metrics: ExportMetricsOption {
@@ -166,8 +167,8 @@ fn test_load_metasrv_example_config() {
                },
                ..Default::default()
            },
-            datanode: meta_srv::metasrv::DatanodeOptions {
-                client: meta_srv::metasrv::DatanodeClientOptions {
+            datanode: DatanodeClientOptions {
+                client: ClientOptions {
                    timeout: Duration::from_secs(10),
                    connect_timeout: Duration::from_secs(10),
                    tcp_nodelay: true,
--- a/src/common/base/Cargo.toml
+++ b/src/common/base/Cargo.toml
@@ -16,9 +16,12 @@ common-error.workspace = true
 common-macro.workspace = true
 futures.workspace = true
 paste = "1.0"
+pin-project.workspace = true
 serde = { version = "1.0", features = ["derive"] }
 snafu.workspace = true
+tokio.workspace = true
 zeroize = { version = "1.6", default-features = false, features = ["alloc"] }

 [dev-dependencies]
+common-test-util.workspace = true
 toml.workspace = true
--- a/src/common/base/src/range_read.rs
+++ b/src/common/base/src/range_read.rs
@@ -12,12 +12,20 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

+use std::future::Future;
 use std::io;
 use std::ops::Range;
+use std::path::Path;
+use std::pin::Pin;
+use std::sync::Arc;
+use std::task::{Context, Poll};

 use async_trait::async_trait;
 use bytes::{BufMut, Bytes};
-use futures::{AsyncReadExt, AsyncSeekExt};
+use futures::AsyncRead;
+use pin_project::pin_project;
+use tokio::io::{AsyncReadExt as _, AsyncSeekExt as _};
+use tokio::sync::Mutex;

 /// `Metadata` contains the metadata of a source.
 pub struct Metadata {
@@ -61,7 +69,7 @@ pub trait RangeReader: Send + Unpin {
 }

 #[async_trait]
-impl<R: RangeReader + Send + Unpin> RangeReader for &mut R {
+impl<R: ?Sized + RangeReader> RangeReader for &mut R {
    async fn metadata(&mut self) -> io::Result<Metadata> {
        (*self).metadata().await
    }
@@ -80,26 +88,212 @@ impl<R: RangeReader + Send + Unpin> RangeReader for &mut R {
    }
 }

-/// `RangeReaderAdapter` bridges `RangeReader` and `AsyncRead + AsyncSeek`.
-pub struct RangeReaderAdapter<R>(pub R);
+/// `AsyncReadAdapter` adapts a `RangeReader` to an `AsyncRead`.
+#[pin_project]
+pub struct AsyncReadAdapter<R> {
+    /// The inner `RangeReader`.
+    /// Use `Mutex` to get rid of the borrow checker issue.
+    inner: Arc<Mutex<R>>,
+
+    /// The current position from the view of the reader.
+    position: u64,
+
+    /// The buffer for the read bytes.
+    buffer: Vec<u8>,
+
+    /// The length of the content.
+    content_length: u64,
+
+    /// The future for reading the next bytes.
+    #[pin]
+    read_fut: Option<Pin<Box<dyn Future<Output = io::Result<Bytes>> + Send>>>,
+}
+
+impl<R: RangeReader + 'static> AsyncReadAdapter<R> {
+    pub async fn new(inner: R) -> io::Result<Self> {
+        let mut inner = inner;
+        let metadata = inner.metadata().await?;
+        Ok(AsyncReadAdapter {
+            inner: Arc::new(Mutex::new(inner)),
+            position: 0,
+            buffer: Vec::new(),
+            content_length: metadata.content_length,
+            read_fut: None,
+        })
+    }
+}
+
+/// The maximum size per read for the inner reader in `AsyncReadAdapter`.
+const MAX_SIZE_PER_READ: usize = 8 * 1024 * 1024; // 8MB
+
+impl<R: RangeReader + 'static> AsyncRead for AsyncReadAdapter<R> {
+    fn poll_read(
+        mut self: Pin<&mut Self>,
+        cx: &mut Context<'_>,
+        buf: &mut [u8],
+    ) -> Poll<io::Result<usize>> {
+        let mut this = self.as_mut().project();
+
+        if *this.position >= *this.content_length {
+            return Poll::Ready(Ok(0));
+        }
+
+        if !this.buffer.is_empty() {
+            let to_read = this.buffer.len().min(buf.len());
+            buf[..to_read].copy_from_slice(&this.buffer[..to_read]);
+            this.buffer.drain(..to_read);
+            *this.position += to_read as u64;
+            return Poll::Ready(Ok(to_read));
+        }
+
+        if this.read_fut.is_none() {
+            let size = (*this.content_length - *this.position).min(MAX_SIZE_PER_READ as u64);
+            let range = *this.position..(*this.position + size);
+            let inner = this.inner.clone();
+            let fut = async move {
+                let mut inner = inner.lock().await;
+                inner.read(range).await
+            };
+
+            *this.read_fut = Some(Box::pin(fut));
+        }
+
+        match this
+            .read_fut
+            .as_mut()
+            .as_pin_mut()
+            .expect("checked above")
+            .poll(cx)
+        {
+            Poll::Pending => Poll::Pending,
+            Poll::Ready(Ok(bytes)) => {
+                *this.read_fut = None;
+
+                if !bytes.is_empty() {
+                    this.buffer.extend_from_slice(&bytes);
+                    self.poll_read(cx, buf)
+                } else {
+                    Poll::Ready(Ok(0))
+                }
+            }
+            Poll::Ready(Err(e)) => {
+                *this.read_fut = None;
+                Poll::Ready(Err(e))
+            }
+        }
+    }
+}

-/// Implements `RangeReader` for a type that implements `AsyncRead + AsyncSeek`.
-///
-/// TODO(zhongzc): It's a temporary solution for porting the codebase from `AsyncRead + AsyncSeek` to `RangeReader`.
-/// Until the codebase is fully ported to `RangeReader`, remove this implementation.
 #[async_trait]
-impl<R: futures::AsyncRead + futures::AsyncSeek + Send + Unpin> RangeReader
-    for RangeReaderAdapter<R>
-{
+impl RangeReader for Vec<u8> {
    async fn metadata(&mut self) -> io::Result<Metadata> {
-        let content_length = self.0.seek(io::SeekFrom::End(0)).await?;
-        Ok(Metadata { content_length })
+        Ok(Metadata {
+            content_length: self.len() as u64,
+        })
    }

-    async fn read(&mut self, range: Range<u64>) -> io::Result<Bytes> {
+    async fn read(&mut self, mut range: Range<u64>) -> io::Result<Bytes> {
+        range.end = range.end.min(self.len() as u64);
+
+        let bytes = Bytes::copy_from_slice(&self[range.start as usize..range.end as usize]);
+        Ok(bytes)
+    }
+}
+
+/// `FileReader` is a `RangeReader` for reading a file.
+pub struct FileReader {
+    content_length: u64,
+    position: u64,
+    file: tokio::fs::File,
+}
+
+impl FileReader {
+    /// Creates a new `FileReader` for the file at the given path.
+    pub async fn new(path: impl AsRef<Path>) -> io::Result<Self> {
+        let file = tokio::fs::File::open(path).await?;
+        let metadata = file.metadata().await?;
+        Ok(FileReader {
+            content_length: metadata.len(),
+            position: 0,
+            file,
+        })
+    }
+}
+
+#[async_trait]
+impl RangeReader for FileReader {
+    async fn metadata(&mut self) -> io::Result<Metadata> {
+        Ok(Metadata {
+            content_length: self.content_length,
+        })
+    }
+
+    async fn read(&mut self, mut range: Range<u64>) -> io::Result<Bytes> {
+        if range.start != self.position {
+            self.file.seek(io::SeekFrom::Start(range.start)).await?;
+            self.position = range.start;
+        }
+
+        range.end = range.end.min(self.content_length);
+        if range.end <= self.position {
+            return Err(io::Error::new(
+                io::ErrorKind::UnexpectedEof,
+                "Start of range is out of bounds",
+            ));
+        }
+
        let mut buf = vec![0; (range.end - range.start) as usize];
-        self.0.seek(io::SeekFrom::Start(range.start)).await?;
-        self.0.read_exact(&mut buf).await?;
+
+        self.file.read_exact(&mut buf).await?;
+        self.position = range.end;
+
        Ok(Bytes::from(buf))
    }
 }
+
+#[cfg(test)]
+mod tests {
+    use common_test_util::temp_dir::create_named_temp_file;
+    use futures::io::AsyncReadExt as _;
+
+    use super::*;
+
+    #[tokio::test]
+    async fn test_async_read_adapter() {
+        let data = b"hello world";
+        let reader = Vec::from(data);
+        let mut adapter = AsyncReadAdapter::new(reader).await.unwrap();
+
+        let mut buf = Vec::new();
+        adapter.read_to_end(&mut buf).await.unwrap();
+        assert_eq!(buf, data);
+    }
+
+    #[tokio::test]
+    async fn test_async_read_adapter_large() {
+        let data = (0..20 * 1024 * 1024).map(|i| i as u8).collect::<Vec<u8>>();
+        let mut adapter = AsyncReadAdapter::new(data.clone()).await.unwrap();
+
+        let mut buf = Vec::new();
+        adapter.read_to_end(&mut buf).await.unwrap();
+        assert_eq!(buf, data);
+    }
+
+    #[tokio::test]
+    async fn test_file_reader() {
+        let file = create_named_temp_file();
+        let path = file.path();
+        let data = b"hello world";
+        tokio::fs::write(path, data).await.unwrap();
+
+        let mut reader = FileReader::new(path).await.unwrap();
+        let metadata = reader.metadata().await.unwrap();
+        assert_eq!(metadata.content_length, data.len() as u64);
+
+        let bytes = reader.read(0..metadata.content_length).await.unwrap();
+        assert_eq!(&*bytes, data);
+
+        let bytes = reader.read(0..5).await.unwrap();
+        assert_eq!(&*bytes, &data[..5]);
+    }
+}
--- a/src/common/function/Cargo.toml
+++ b/src/common/function/Cargo.toml
@@ -9,7 +9,7 @@ workspace = true

 [features]
 default = ["geo"]
-geo = ["geohash", "h3o", "s2"]
+geo = ["geohash", "h3o", "s2", "wkt", "geo-types", "dep:geo"]

 [dependencies]
 api.workspace = true
@@ -28,6 +28,8 @@ common-version.workspace = true
 datafusion.workspace = true
 datatypes.workspace = true
 derive_more = { version = "1", default-features = false, features = ["display"] }
+geo = { version = "0.29", optional = true }
+geo-types = { version = "0.7", optional = true }
 geohash = { version = "0.13", optional = true }
 h3o = { version = "0.6", optional = true }
 jsonb.workspace = true
@@ -44,6 +46,7 @@ sql.workspace = true
 statrs = "0.16"
 store-api.workspace = true
 table.workspace = true
+wkt = { version = "0.11", optional = true }

 [dev-dependencies]
 ron = "0.7"
--- a/src/common/function/src/scalars/geo.rs
+++ b/src/common/function/src/scalars/geo.rs
@@ -17,7 +17,10 @@ pub(crate) mod encoding;
 mod geohash;
 mod h3;
 mod helpers;
+mod measure;
+mod relation;
 mod s2;
+mod wkt;

 use crate::function_registry::FunctionRegistry;

@@ -48,6 +51,7 @@ impl GeoFunctions {
        registry.register(Arc::new(h3::H3CellToChildrenSize));
        registry.register(Arc::new(h3::H3CellToChildPos));
        registry.register(Arc::new(h3::H3ChildPosToCell));
+        registry.register(Arc::new(h3::H3CellContains));

        // h3 grid traversal
        registry.register(Arc::new(h3::H3GridDisk));
@@ -55,10 +59,27 @@ impl GeoFunctions {
        registry.register(Arc::new(h3::H3GridDistance));
        registry.register(Arc::new(h3::H3GridPathCells));

+        // h3 measurement
+        registry.register(Arc::new(h3::H3CellDistanceSphereKm));
+        registry.register(Arc::new(h3::H3CellDistanceEuclideanDegree));
+
        // s2
        registry.register(Arc::new(s2::S2LatLngToCell));
        registry.register(Arc::new(s2::S2CellLevel));
        registry.register(Arc::new(s2::S2CellToToken));
        registry.register(Arc::new(s2::S2CellParent));
+
+        // spatial data type
+        registry.register(Arc::new(wkt::LatLngToPointWkt));
+
+        // spatial relation
+        registry.register(Arc::new(relation::STContains));
+        registry.register(Arc::new(relation::STWithin));
+        registry.register(Arc::new(relation::STIntersects));
+
+        // spatial measure
+        registry.register(Arc::new(measure::STDistance));
+        registry.register(Arc::new(measure::STDistanceSphere));
+        registry.register(Arc::new(measure::STArea));
    }
 }
--- a/src/common/function/src/scalars/geo/h3.rs
+++ b/src/common/function/src/scalars/geo/h3.rs
@@ -23,8 +23,8 @@ use datatypes::prelude::ConcreteDataType;
 use datatypes::scalars::{Scalar, ScalarVectorBuilder};
 use datatypes::value::{ListValue, Value};
 use datatypes::vectors::{
-    BooleanVectorBuilder, Int32VectorBuilder, ListVectorBuilder, MutableVector,
-    StringVectorBuilder, UInt64VectorBuilder, UInt8VectorBuilder, VectorRef,
+    BooleanVectorBuilder, Float64VectorBuilder, Int32VectorBuilder, ListVectorBuilder,
+    MutableVector, StringVectorBuilder, UInt64VectorBuilder, UInt8VectorBuilder, VectorRef,
 };
 use derive_more::Display;
 use h3o::{CellIndex, LatLng, Resolution};
@@ -38,6 +38,7 @@ static CELL_TYPES: Lazy<Vec<ConcreteDataType>> = Lazy::new(|| {
    vec![
        ConcreteDataType::int64_datatype(),
        ConcreteDataType::uint64_datatype(),
+        ConcreteDataType::string_datatype(),
    ]
 });

@@ -952,6 +953,181 @@ impl Function for H3GridPathCells {
    }
 }

+/// Tests if cells contains given cells
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct H3CellContains;
+
+impl Function for H3CellContains {
+    fn name(&self) -> &str {
+        "h3_cells_contains"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::boolean_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        let multi_cell_types = vec![
+            ConcreteDataType::list_datatype(ConcreteDataType::int64_datatype()),
+            ConcreteDataType::list_datatype(ConcreteDataType::uint64_datatype()),
+            ConcreteDataType::list_datatype(ConcreteDataType::string_datatype()),
+            ConcreteDataType::string_datatype(),
+        ];
+
+        let mut signatures = Vec::with_capacity(multi_cell_types.len() * CELL_TYPES.len());
+        for multi_cell_type in &multi_cell_types {
+            for cell_type in CELL_TYPES.as_slice() {
+                signatures.push(TypeSignature::Exact(vec![
+                    multi_cell_type.clone(),
+                    cell_type.clone(),
+                ]));
+            }
+        }
+
+        Signature::one_of(signatures, Volatility::Stable)
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let cells_vec = &columns[0];
+        let cell_this_vec = &columns[1];
+
+        let size = cell_this_vec.len();
+        let mut results = BooleanVectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let mut result = None;
+            if let (cells, Some(cell_this)) = (
+                cells_from_value(cells_vec.get(i))?,
+                cell_from_value(cell_this_vec.get(i))?,
+            ) {
+                result = Some(false);
+
+                for cell_that in cells.iter() {
+                    // get cell resolution, and find cell_this's parent at
+                    //  this solution, test if cell_that equals the parent
+                    let resolution = cell_that.resolution();
+                    if let Some(cell_this_parent) = cell_this.parent(resolution) {
+                        if cell_this_parent == *cell_that {
+                            result = Some(true);
+                            break;
+                        }
+                    }
+                }
+            }
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+/// Get WGS84 great circle distance of two cell centroid
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct H3CellDistanceSphereKm;
+
+impl Function for H3CellDistanceSphereKm {
+    fn name(&self) -> &str {
+        "h3_distance_sphere_km"
+    }
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::float64_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        signature_of_double_cells()
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let cell_this_vec = &columns[0];
+        let cell_that_vec = &columns[1];
+        let size = cell_this_vec.len();
+
+        let mut results = Float64VectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let result = match (
+                cell_from_value(cell_this_vec.get(i))?,
+                cell_from_value(cell_that_vec.get(i))?,
+            ) {
+                (Some(cell_this), Some(cell_that)) => {
+                    let centroid_this = LatLng::from(cell_this);
+                    let centroid_that = LatLng::from(cell_that);
+
+                    Some(centroid_this.distance_km(centroid_that))
+                }
+                _ => None,
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+/// Get Euclidean distance of two cell centroid
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct H3CellDistanceEuclideanDegree;
+
+impl H3CellDistanceEuclideanDegree {
+    fn distance(centroid_this: LatLng, centroid_that: LatLng) -> f64 {
+        ((centroid_this.lat() - centroid_that.lat()).powi(2)
+            + (centroid_this.lng() - centroid_that.lng()).powi(2))
+        .sqrt()
+    }
+}
+
+impl Function for H3CellDistanceEuclideanDegree {
+    fn name(&self) -> &str {
+        "h3_distance_degree"
+    }
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::float64_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        signature_of_double_cells()
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let cell_this_vec = &columns[0];
+        let cell_that_vec = &columns[1];
+        let size = cell_this_vec.len();
+
+        let mut results = Float64VectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let result = match (
+                cell_from_value(cell_this_vec.get(i))?,
+                cell_from_value(cell_that_vec.get(i))?,
+            ) {
+                (Some(cell_this), Some(cell_that)) => {
+                    let centroid_this = LatLng::from(cell_this);
+                    let centroid_that = LatLng::from(cell_that);
+
+                    let dist = Self::distance(centroid_this, centroid_that);
+                    Some(dist)
+                }
+                _ => None,
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
 fn value_to_resolution(v: Value) -> Result<Resolution> {
    let r = match v {
        Value::Int8(v) => v as u8,
@@ -1073,7 +1249,126 @@ fn cell_from_value(v: Value) -> Result<Option<CellIndex>> {
                })
                .context(error::ExecuteSnafu)?,
        ),
+        Value::String(s) => Some(
+            CellIndex::from_str(s.as_utf8())
+                .map_err(|e| {
+                    BoxedError::new(PlainError::new(
+                        format!("H3 error: {}", e),
+                        StatusCode::EngineExecuteQuery,
+                    ))
+                })
+                .context(error::ExecuteSnafu)?,
+        ),
        _ => None,
    };
    Ok(cell)
 }
+
+/// extract cell array from all possible types including:
+/// - int64 list
+/// - uint64 list
+/// - string list
+/// - comma-separated string
+fn cells_from_value(v: Value) -> Result<Vec<CellIndex>> {
+    match v {
+        Value::List(list) => match list.datatype() {
+            ConcreteDataType::Int64(_) => list
+                .items()
+                .iter()
+                .map(|v| {
+                    if let Value::Int64(v) = v {
+                        CellIndex::try_from(*v as u64)
+                            .map_err(|e| {
+                                BoxedError::new(PlainError::new(
+                                    format!("H3 error: {}", e),
+                                    StatusCode::EngineExecuteQuery,
+                                ))
+                            })
+                            .context(error::ExecuteSnafu)
+                    } else {
+                        Err(BoxedError::new(PlainError::new(
+                            "Invalid data type in array".to_string(),
+                            StatusCode::EngineExecuteQuery,
+                        )))
+                        .context(error::ExecuteSnafu)
+                    }
+                })
+                .collect::<Result<Vec<CellIndex>>>(),
+            ConcreteDataType::UInt64(_) => list
+                .items()
+                .iter()
+                .map(|v| {
+                    if let Value::UInt64(v) = v {
+                        CellIndex::try_from(*v)
+                            .map_err(|e| {
+                                BoxedError::new(PlainError::new(
+                                    format!("H3 error: {}", e),
+                                    StatusCode::EngineExecuteQuery,
+                                ))
+                            })
+                            .context(error::ExecuteSnafu)
+                    } else {
+                        Err(BoxedError::new(PlainError::new(
+                            "Invalid data type in array".to_string(),
+                            StatusCode::EngineExecuteQuery,
+                        )))
+                        .context(error::ExecuteSnafu)
+                    }
+                })
+                .collect::<Result<Vec<CellIndex>>>(),
+            ConcreteDataType::String(_) => list
+                .items()
+                .iter()
+                .map(|v| {
+                    if let Value::String(v) = v {
+                        CellIndex::from_str(v.as_utf8().trim())
+                            .map_err(|e| {
+                                BoxedError::new(PlainError::new(
+                                    format!("H3 error: {}", e),
+                                    StatusCode::EngineExecuteQuery,
+                                ))
+                            })
+                            .context(error::ExecuteSnafu)
+                    } else {
+                        Err(BoxedError::new(PlainError::new(
+                            "Invalid data type in array".to_string(),
+                            StatusCode::EngineExecuteQuery,
+                        )))
+                        .context(error::ExecuteSnafu)
+                    }
+                })
+                .collect::<Result<Vec<CellIndex>>>(),
+            _ => Ok(vec![]),
+        },
+        Value::String(csv) => {
+            let str_seq = csv.as_utf8().split(',');
+            str_seq
+                .map(|v| {
+                    CellIndex::from_str(v.trim())
+                        .map_err(|e| {
+                            BoxedError::new(PlainError::new(
+                                format!("H3 error: {}", e),
+                                StatusCode::EngineExecuteQuery,
+                            ))
+                        })
+                        .context(error::ExecuteSnafu)
+                })
+                .collect::<Result<Vec<CellIndex>>>()
+        }
+        _ => Ok(vec![]),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_h3_euclidean_distance() {
+        let point_this = LatLng::new(42.3521, -72.1235).expect("incorrect lat lng");
+        let point_that = LatLng::new(42.45, -72.1260).expect("incorrect lat lng");
+
+        let dist = H3CellDistanceEuclideanDegree::distance(point_this, point_that);
+        assert_eq!(dist, 0.09793191512474639);
+    }
+}
--- a/src/common/function/src/scalars/geo/measure.rs
+++ b/src/common/function/src/scalars/geo/measure.rs
@@ -0,0 +1,195 @@
+// 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 common_error::ext::{BoxedError, PlainError};
+use common_error::status_code::StatusCode;
+use common_query::error::{self, Result};
+use common_query::prelude::{Signature, TypeSignature};
+use datafusion::logical_expr::Volatility;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{Float64VectorBuilder, MutableVector, VectorRef};
+use derive_more::Display;
+use geo::algorithm::line_measures::metric_spaces::Euclidean;
+use geo::{Area, Distance, Haversine};
+use geo_types::Geometry;
+use snafu::ResultExt;
+
+use super::helpers::{ensure_columns_len, ensure_columns_n};
+use super::wkt::parse_wkt;
+use crate::function::{Function, FunctionContext};
+
+/// Return WGS84(SRID: 4326) euclidean distance between two geometry object, in degree
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct STDistance;
+
+impl Function for STDistance {
+    fn name(&self) -> &str {
+        "st_distance"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::float64_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::new(
+            TypeSignature::Exact(vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::string_datatype(),
+            ]),
+            Volatility::Stable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let wkt_this_vec = &columns[0];
+        let wkt_that_vec = &columns[1];
+
+        let size = wkt_this_vec.len();
+        let mut results = Float64VectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let wkt_this = wkt_this_vec.get(i).as_string();
+            let wkt_that = wkt_that_vec.get(i).as_string();
+
+            let result = match (wkt_this, wkt_that) {
+                (Some(wkt_this), Some(wkt_that)) => {
+                    let geom_this = parse_wkt(&wkt_this)?;
+                    let geom_that = parse_wkt(&wkt_that)?;
+
+                    Some(Euclidean::distance(&geom_this, &geom_that))
+                }
+                _ => None,
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+/// Return great circle distance between two geometry object, in meters
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct STDistanceSphere;
+
+impl Function for STDistanceSphere {
+    fn name(&self) -> &str {
+        "st_distance_sphere_m"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::float64_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::new(
+            TypeSignature::Exact(vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::string_datatype(),
+            ]),
+            Volatility::Stable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let wkt_this_vec = &columns[0];
+        let wkt_that_vec = &columns[1];
+
+        let size = wkt_this_vec.len();
+        let mut results = Float64VectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let wkt_this = wkt_this_vec.get(i).as_string();
+            let wkt_that = wkt_that_vec.get(i).as_string();
+
+            let result = match (wkt_this, wkt_that) {
+                (Some(wkt_this), Some(wkt_that)) => {
+                    let geom_this = parse_wkt(&wkt_this)?;
+                    let geom_that = parse_wkt(&wkt_that)?;
+
+                    match (geom_this, geom_that) {
+                        (Geometry::Point(this), Geometry::Point(that)) => {
+                            Some(Haversine::distance(this, that))
+                        }
+                        _ => {
+                            Err(BoxedError::new(PlainError::new(
+                                "Great circle distance between non-point objects are not supported for now.".to_string(),
+                                StatusCode::Unsupported,
+                            ))).context(error::ExecuteSnafu)?
+                        }
+                    }
+                }
+                _ => None,
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+/// Return area of given geometry object
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct STArea;
+
+impl Function for STArea {
+    fn name(&self) -> &str {
+        "st_area"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::float64_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::new(
+            TypeSignature::Exact(vec![ConcreteDataType::string_datatype()]),
+            Volatility::Stable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 1);
+
+        let wkt_vec = &columns[0];
+
+        let size = wkt_vec.len();
+        let mut results = Float64VectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let wkt = wkt_vec.get(i).as_string();
+
+            let result = if let Some(wkt) = wkt {
+                let geom = parse_wkt(&wkt)?;
+                Some(geom.unsigned_area())
+            } else {
+                None
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
--- a/src/common/function/src/scalars/geo/relation.rs
+++ b/src/common/function/src/scalars/geo/relation.rs
@@ -0,0 +1,190 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use common_query::error::Result;
+use common_query::prelude::{Signature, TypeSignature};
+use datafusion::logical_expr::Volatility;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{BooleanVectorBuilder, MutableVector, VectorRef};
+use derive_more::Display;
+use geo::algorithm::contains::Contains;
+use geo::algorithm::intersects::Intersects;
+use geo::algorithm::within::Within;
+
+use super::helpers::{ensure_columns_len, ensure_columns_n};
+use super::wkt::parse_wkt;
+use crate::function::{Function, FunctionContext};
+
+/// Test if spatial relationship: contains
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct STContains;
+
+impl Function for STContains {
+    fn name(&self) -> &str {
+        "st_contains"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::boolean_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::new(
+            TypeSignature::Exact(vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::string_datatype(),
+            ]),
+            Volatility::Stable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let wkt_this_vec = &columns[0];
+        let wkt_that_vec = &columns[1];
+
+        let size = wkt_this_vec.len();
+        let mut results = BooleanVectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let wkt_this = wkt_this_vec.get(i).as_string();
+            let wkt_that = wkt_that_vec.get(i).as_string();
+
+            let result = match (wkt_this, wkt_that) {
+                (Some(wkt_this), Some(wkt_that)) => {
+                    let geom_this = parse_wkt(&wkt_this)?;
+                    let geom_that = parse_wkt(&wkt_that)?;
+
+                    Some(geom_this.contains(&geom_that))
+                }
+                _ => None,
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+/// Test if spatial relationship: within
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct STWithin;
+
+impl Function for STWithin {
+    fn name(&self) -> &str {
+        "st_within"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::boolean_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::new(
+            TypeSignature::Exact(vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::string_datatype(),
+            ]),
+            Volatility::Stable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let wkt_this_vec = &columns[0];
+        let wkt_that_vec = &columns[1];
+
+        let size = wkt_this_vec.len();
+        let mut results = BooleanVectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let wkt_this = wkt_this_vec.get(i).as_string();
+            let wkt_that = wkt_that_vec.get(i).as_string();
+
+            let result = match (wkt_this, wkt_that) {
+                (Some(wkt_this), Some(wkt_that)) => {
+                    let geom_this = parse_wkt(&wkt_this)?;
+                    let geom_that = parse_wkt(&wkt_that)?;
+
+                    Some(geom_this.is_within(&geom_that))
+                }
+                _ => None,
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+/// Test if spatial relationship: within
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct STIntersects;
+
+impl Function for STIntersects {
+    fn name(&self) -> &str {
+        "st_intersects"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::boolean_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::new(
+            TypeSignature::Exact(vec![
+                ConcreteDataType::string_datatype(),
+                ConcreteDataType::string_datatype(),
+            ]),
+            Volatility::Stable,
+        )
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let wkt_this_vec = &columns[0];
+        let wkt_that_vec = &columns[1];
+
+        let size = wkt_this_vec.len();
+        let mut results = BooleanVectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let wkt_this = wkt_this_vec.get(i).as_string();
+            let wkt_that = wkt_that_vec.get(i).as_string();
+
+            let result = match (wkt_this, wkt_that) {
+                (Some(wkt_this), Some(wkt_that)) => {
+                    let geom_this = parse_wkt(&wkt_this)?;
+                    let geom_that = parse_wkt(&wkt_that)?;
+
+                    Some(geom_this.intersects(&geom_that))
+                }
+                _ => None,
+            };
+
+            results.push(result);
+        }
+
+        Ok(results.to_vector())
+    }
+}
--- a/src/common/function/src/scalars/geo/wkt.rs
+++ b/src/common/function/src/scalars/geo/wkt.rs
@@ -0,0 +1,100 @@
+// 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 common_error::ext::{BoxedError, PlainError};
+use common_error::status_code::StatusCode;
+use common_query::error::{self, Result};
+use common_query::prelude::{Signature, TypeSignature};
+use datafusion::logical_expr::Volatility;
+use datatypes::prelude::ConcreteDataType;
+use datatypes::scalars::ScalarVectorBuilder;
+use datatypes::vectors::{MutableVector, StringVectorBuilder, VectorRef};
+use derive_more::Display;
+use geo_types::{Geometry, Point};
+use once_cell::sync::Lazy;
+use snafu::ResultExt;
+use wkt::{ToWkt, TryFromWkt};
+
+use super::helpers::{ensure_columns_len, ensure_columns_n};
+use crate::function::{Function, FunctionContext};
+
+static COORDINATE_TYPES: Lazy<Vec<ConcreteDataType>> = Lazy::new(|| {
+    vec![
+        ConcreteDataType::float32_datatype(),
+        ConcreteDataType::float64_datatype(),
+    ]
+});
+
+/// Return WGS84(SRID: 4326) euclidean distance between two geometry object, in degree
+#[derive(Clone, Debug, Default, Display)]
+#[display("{}", self.name())]
+pub struct LatLngToPointWkt;
+
+impl Function for LatLngToPointWkt {
+    fn name(&self) -> &str {
+        "wkt_point_from_latlng"
+    }
+
+    fn return_type(&self, _input_types: &[ConcreteDataType]) -> Result<ConcreteDataType> {
+        Ok(ConcreteDataType::string_datatype())
+    }
+
+    fn signature(&self) -> Signature {
+        let mut signatures = Vec::new();
+        for coord_type in COORDINATE_TYPES.as_slice() {
+            signatures.push(TypeSignature::Exact(vec![
+                // latitude
+                coord_type.clone(),
+                // longitude
+                coord_type.clone(),
+            ]));
+        }
+        Signature::one_of(signatures, Volatility::Stable)
+    }
+
+    fn eval(&self, _func_ctx: FunctionContext, columns: &[VectorRef]) -> Result<VectorRef> {
+        ensure_columns_n!(columns, 2);
+
+        let lat_vec = &columns[0];
+        let lng_vec = &columns[1];
+
+        let size = lat_vec.len();
+        let mut results = StringVectorBuilder::with_capacity(size);
+
+        for i in 0..size {
+            let lat = lat_vec.get(i).as_f64_lossy();
+            let lng = lng_vec.get(i).as_f64_lossy();
+
+            let result = match (lat, lng) {
+                (Some(lat), Some(lng)) => Some(Point::new(lng, lat).wkt_string()),
+                _ => None,
+            };
+
+            results.push(result.as_deref());
+        }
+
+        Ok(results.to_vector())
+    }
+}
+
+pub(super) fn parse_wkt(s: &str) -> Result<Geometry> {
+    Geometry::try_from_wkt_str(s)
+        .map_err(|e| {
+            BoxedError::new(PlainError::new(
+                format!("Fail to parse WKT: {}", e),
+                StatusCode::EngineExecuteQuery,
+            ))
+        })
+        .context(error::ExecuteSnafu)
+}
--- a/src/common/function/src/scalars/json/json_path_exists.rs
+++ b/src/common/function/src/scalars/json/json_path_exists.rs
@@ -15,7 +15,7 @@
 use std::fmt::{self, Display};

 use common_query::error::{InvalidFuncArgsSnafu, Result, UnsupportedInputDataTypeSnafu};
-use common_query::prelude::Signature;
+use common_query::prelude::{Signature, TypeSignature};
 use datafusion::logical_expr::Volatility;
 use datatypes::data_type::ConcreteDataType;
 use datatypes::prelude::VectorRef;
@@ -41,10 +41,24 @@ impl Function for JsonPathExistsFunction {
    }

    fn signature(&self) -> Signature {
-        Signature::exact(
+        Signature::one_of(
            vec![
-                ConcreteDataType::json_datatype(),
-                ConcreteDataType::string_datatype(),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::string_datatype(),
+                ]),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::null_datatype(),
+                    ConcreteDataType::string_datatype(),
+                ]),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::null_datatype(),
+                ]),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::null_datatype(),
+                    ConcreteDataType::null_datatype(),
+                ]),
            ],
            Volatility::Immutable,
        )
@@ -64,25 +78,26 @@ impl Function for JsonPathExistsFunction {
        let paths = &columns[1];

        let size = jsons.len();
-        let datatype = jsons.data_type();
        let mut results = BooleanVectorBuilder::with_capacity(size);

-        match datatype {
-            // JSON data type uses binary vector
-            ConcreteDataType::Binary(_) => {
+        match (jsons.data_type(), paths.data_type()) {
+            (ConcreteDataType::Binary(_), ConcreteDataType::String(_)) => {
                for i in 0..size {
-                    let json = jsons.get_ref(i);
-                    let path = paths.get_ref(i);
-
-                    let json = json.as_binary();
-                    let path = path.as_string();
-                    let result = match (json, path) {
+                    let result = match (jsons.get_ref(i).as_binary(), paths.get_ref(i).as_string())
+                    {
                        (Ok(Some(json)), Ok(Some(path))) => {
-                            let json_path = jsonb::jsonpath::parse_json_path(path.as_bytes());
-                            match json_path {
-                                Ok(json_path) => jsonb::path_exists(json, json_path).ok(),
-                                Err(_) => None,
-                            }
+                            // Get `JsonPath`.
+                            let json_path = match jsonb::jsonpath::parse_json_path(path.as_bytes())
+                            {
+                                Ok(json_path) => json_path,
+                                Err(_) => {
+                                    return InvalidFuncArgsSnafu {
+                                        err_msg: format!("Illegal json path: {:?}", path),
+                                    }
+                                    .fail();
+                                }
+                            };
+                            jsonb::path_exists(json, json_path).ok()
                        }
                        _ => None,
                    };
@@ -90,6 +105,12 @@ impl Function for JsonPathExistsFunction {
                    results.push(result);
                }
            }
+
+            // Any null args existence causes the result to be NULL.
+            (ConcreteDataType::Null(_), ConcreteDataType::String(_)) => results.push_nulls(size),
+            (ConcreteDataType::Binary(_), ConcreteDataType::Null(_)) => results.push_nulls(size),
+            (ConcreteDataType::Null(_), ConcreteDataType::Null(_)) => results.push_nulls(size),
+
            _ => {
                return UnsupportedInputDataTypeSnafu {
                    function: NAME,
@@ -114,8 +135,8 @@ mod tests {
    use std::sync::Arc;

    use common_query::prelude::TypeSignature;
-    use datatypes::scalars::ScalarVector;
-    use datatypes::vectors::{BinaryVector, StringVector};
+    use datatypes::prelude::ScalarVector;
+    use datatypes::vectors::{BinaryVector, NullVector, StringVector};

    use super::*;

@@ -133,9 +154,27 @@ mod tests {

        assert!(matches!(json_path_exists.signature(),
                         Signature {
-                             type_signature: TypeSignature::Exact(valid_types),
+                             type_signature: TypeSignature::OneOf(valid_types),
                             volatility: Volatility::Immutable
-                         } if  valid_types == vec![ConcreteDataType::json_datatype(), ConcreteDataType::string_datatype()]
+                         } if valid_types ==
+            vec![
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::string_datatype(),
+                ]),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::null_datatype(),
+                    ConcreteDataType::string_datatype(),
+                ]),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::json_datatype(),
+                    ConcreteDataType::null_datatype(),
+                ]),
+                TypeSignature::Exact(vec![
+                    ConcreteDataType::null_datatype(),
+                    ConcreteDataType::null_datatype(),
+                ]),
+            ],
        ));

        let json_strings = [
@@ -143,9 +182,15 @@ mod tests {
            r#"{"a": 4, "b": {"c": 6}, "c": 6}"#,
            r#"{"a": 7, "b": 8, "c": {"a": 7}}"#,
            r#"{"a": 7, "b": 8, "c": {"a": 7}}"#,
+            r#"[1, 2, 3]"#,
+            r#"null"#,
+            r#"{"a": 7, "b": 8, "c": {"a": 7}}"#,
+            r#"null"#,
        ];
-        let paths = vec!["$.a.b.c", "$.b", "$.c.a", ".d"];
-        let results = [false, true, true, false];
+        let paths = vec![
+            "$.a.b.c", "$.b", "$.c.a", ".d", "$[0]", "$.a", "null", "null",
+        ];
+        let expected = [false, true, true, false, true, false, false, false];

        let jsonbs = json_strings
            .iter()
@@ -162,11 +207,44 @@ mod tests {
            .eval(FunctionContext::default(), &args)
            .unwrap();

-        assert_eq!(4, vector.len());
-        for (i, gt) in results.iter().enumerate() {
+        // Test for non-nulls.
+        assert_eq!(8, vector.len());
+        for (i, real) in expected.iter().enumerate() {
            let result = vector.get_ref(i);
-            let result = result.as_boolean().unwrap().unwrap();
-            assert_eq!(*gt, result);
+            assert!(!result.is_null());
+            let val = result.as_boolean().unwrap().unwrap();
+            assert_eq!(val, *real);
        }
+
+        // Test for path error.
+        let json_bytes = jsonb::parse_value("{}".as_bytes()).unwrap().to_vec();
+        let json = BinaryVector::from_vec(vec![json_bytes]);
+        let illegal_path = StringVector::from_vec(vec!["$..a"]);
+
+        let args: Vec<VectorRef> = vec![Arc::new(json), Arc::new(illegal_path)];
+        let err = json_path_exists.eval(FunctionContext::default(), &args);
+        assert!(err.is_err());
+
+        // Test for nulls.
+        let json_bytes = jsonb::parse_value("{}".as_bytes()).unwrap().to_vec();
+        let json = BinaryVector::from_vec(vec![json_bytes]);
+        let null_json = NullVector::new(1);
+
+        let path = StringVector::from_vec(vec!["$.a"]);
+        let null_path = NullVector::new(1);
+
+        let args: Vec<VectorRef> = vec![Arc::new(null_json), Arc::new(path)];
+        let result1 = json_path_exists
+            .eval(FunctionContext::default(), &args)
+            .unwrap();
+        let args: Vec<VectorRef> = vec![Arc::new(json), Arc::new(null_path)];
+        let result2 = json_path_exists
+            .eval(FunctionContext::default(), &args)
+            .unwrap();
+
+        assert_eq!(result1.len(), 1);
+        assert!(result1.get_ref(0).is_null());
+        assert_eq!(result2.len(), 1);
+        assert!(result2.get_ref(0).is_null());
    }
 }
--- a/src/common/grpc-expr/src/alter.rs
+++ b/src/common/grpc-expr/src/alter.rs
@@ -15,20 +15,22 @@
 use api::helper::ColumnDataTypeWrapper;
 use api::v1::add_column_location::LocationType;
 use api::v1::alter_expr::Kind;
+use api::v1::column_def::as_fulltext_option;
 use api::v1::{
-    column_def, AddColumnLocation as Location, AlterExpr, ChangeColumnTypes, CreateTableExpr,
-    DropColumns, RenameTable, SemanticType,
+    column_def, AddColumnLocation as Location, AlterExpr, Analyzer, ChangeColumnTypes,
+    CreateTableExpr, DropColumns, RenameTable, SemanticType,
 };
 use common_query::AddColumnLocation;
-use datatypes::schema::{ColumnSchema, RawSchema};
+use datatypes::schema::{ColumnSchema, FulltextOptions, RawSchema};
 use snafu::{ensure, OptionExt, ResultExt};
 use store_api::region_request::ChangeOption;
 use table::metadata::TableId;
 use table::requests::{AddColumnRequest, AlterKind, AlterTableRequest, ChangeColumnTypeRequest};

 use crate::error::{
-    InvalidChangeTableOptionRequestSnafu, InvalidColumnDefSnafu, MissingFieldSnafu,
-    MissingTimestampColumnSnafu, Result, UnknownLocationTypeSnafu,
+    InvalidChangeFulltextOptionRequestSnafu, InvalidChangeTableOptionRequestSnafu,
+    InvalidColumnDefSnafu, MissingFieldSnafu, MissingTimestampColumnSnafu, Result,
+    UnknownLocationTypeSnafu,
 };

 const LOCATION_TYPE_FIRST: i32 = LocationType::First as i32;
@@ -102,6 +104,17 @@ pub fn alter_expr_to_request(table_id: TableId, expr: AlterExpr) -> Result<Alter
                .collect::<std::result::Result<Vec<_>, _>>()
                .context(InvalidChangeTableOptionRequestSnafu)?,
        },
+        Kind::ChangeColumnFulltext(c) => AlterKind::ChangeColumnFulltext {
+            column_name: c.column_name,
+            options: FulltextOptions {
+                enable: c.enable,
+                analyzer: as_fulltext_option(
+                    Analyzer::try_from(c.analyzer)
+                        .context(InvalidChangeFulltextOptionRequestSnafu)?,
+                ),
+                case_sensitive: c.case_sensitive,
+            },
+        },
    };

    let request = AlterTableRequest {
--- a/src/common/grpc-expr/src/error.rs
+++ b/src/common/grpc-expr/src/error.rs
@@ -125,6 +125,14 @@ pub enum Error {
        #[snafu(source)]
        error: MetadataError,
    },
+
+    #[snafu(display("Invalid change fulltext option request"))]
+    InvalidChangeFulltextOptionRequest {
+        #[snafu(implicit)]
+        location: Location,
+        #[snafu(source)]
+        error: prost::DecodeError,
+    },
 }

 pub type Result<T> = std::result::Result<T, Error>;
@@ -148,7 +156,8 @@ impl ErrorExt for Error {
            Error::UnknownColumnDataType { .. } | Error::InvalidFulltextColumnType { .. } => {
                StatusCode::InvalidArguments
            }
-            Error::InvalidChangeTableOptionRequest { .. } => StatusCode::InvalidArguments,
+            Error::InvalidChangeTableOptionRequest { .. }
+            | Error::InvalidChangeFulltextOptionRequest { .. } => StatusCode::InvalidArguments,
        }
    }

--- a/src/common/grpc-expr/src/util.rs
+++ b/src/common/grpc-expr/src/util.rs
@@ -14,10 +14,11 @@

 use std::collections::HashSet;

+use api::v1::column_data_type_extension::TypeExt;
 use api::v1::column_def::contains_fulltext;
 use api::v1::{
    AddColumn, AddColumns, Column, ColumnDataType, ColumnDataTypeExtension, ColumnDef,
-    ColumnOptions, ColumnSchema, CreateTableExpr, SemanticType,
+    ColumnOptions, ColumnSchema, CreateTableExpr, JsonTypeExtension, SemanticType,
 };
 use datatypes::schema::Schema;
 use snafu::{ensure, OptionExt, ResultExt};
@@ -25,8 +26,9 @@ use table::metadata::TableId;
 use table::table_reference::TableReference;

 use crate::error::{
-    DuplicatedColumnNameSnafu, DuplicatedTimestampColumnSnafu, InvalidFulltextColumnTypeSnafu,
-    MissingTimestampColumnSnafu, Result, UnknownColumnDataTypeSnafu,
+    self, DuplicatedColumnNameSnafu, DuplicatedTimestampColumnSnafu,
+    InvalidFulltextColumnTypeSnafu, MissingTimestampColumnSnafu, Result,
+    UnknownColumnDataTypeSnafu,
 };
 pub struct ColumnExpr<'a> {
    pub column_name: &'a str,
@@ -72,6 +74,28 @@ impl<'a> From<&'a ColumnSchema> for ColumnExpr<'a> {
    }
 }

+fn infer_column_datatype(
+    datatype: i32,
+    datatype_extension: &Option<ColumnDataTypeExtension>,
+) -> Result<ColumnDataType> {
+    let column_type =
+        ColumnDataType::try_from(datatype).context(UnknownColumnDataTypeSnafu { datatype })?;
+
+    if matches!(&column_type, ColumnDataType::Binary) {
+        if let Some(ext) = datatype_extension {
+            let type_ext = ext
+                .type_ext
+                .as_ref()
+                .context(error::MissingFieldSnafu { field: "type_ext" })?;
+            if *type_ext == TypeExt::JsonType(JsonTypeExtension::JsonBinary.into()) {
+                return Ok(ColumnDataType::Json);
+            }
+        }
+    }
+
+    Ok(column_type)
+}
+
 pub fn build_create_table_expr(
    table_id: Option<TableId>,
    table_name: &TableReference<'_>,
@@ -124,8 +148,7 @@ pub fn build_create_table_expr(
            _ => {}
        }

-        let column_type =
-            ColumnDataType::try_from(datatype).context(UnknownColumnDataTypeSnafu { datatype })?;
+        let column_type = infer_column_datatype(datatype, datatype_extension)?;

        ensure!(
            !contains_fulltext(options) || column_type == ColumnDataType::String,
--- a/src/common/grpc/src/select.rs
+++ b/src/common/grpc/src/select.rs
@@ -218,6 +218,12 @@ pub fn values(arrays: &[VectorRef]) -> Result<Values> {
            Decimal128Vector,
            decimal128_values,
            |x| { convert_to_pb_decimal128(x) }
+        ),
+        (
+            ConcreteDataType::Vector(_),
+            BinaryVector,
+            binary_values,
+            |x| { x.into() }
        )
    )
 }
--- a/src/common/meta/Cargo.toml
+++ b/src/common/meta/Cargo.toml
@@ -39,7 +39,7 @@ derive_builder.workspace = true
 etcd-client.workspace = true
 futures.workspace = true
 futures-util.workspace = true
-hex = { version = "0.4" }
+hex.workspace = true
 humantime-serde.workspace = true
 itertools.workspace = true
 lazy_static.workspace = true
--- a/src/common/meta/src/datanode.rs
+++ b/src/common/meta/src/datanode.rs
@@ -77,7 +77,7 @@ pub struct RegionStat {
    pub rcus: i64,
    /// The write capacity units during this period
    pub wcus: i64,
-    /// Approximate bytes of this region
+    /// Approximate disk bytes of this region, including sst, index, manifest and wal
    pub approximate_bytes: u64,
    /// The engine name.
    pub engine: String,
--- a/src/common/meta/src/ddl/alter_table/region_request.rs
+++ b/src/common/meta/src/ddl/alter_table/region_request.rs
@@ -107,6 +107,9 @@ fn create_proto_alter_kind(
        }
        Kind::RenameTable(_) => Ok(None),
        Kind::ChangeTableOptions(v) => Ok(Some(alter_request::Kind::ChangeTableOptions(v.clone()))),
+        Kind::ChangeColumnFulltext(v) => {
+            Ok(Some(alter_request::Kind::ChangeColumnFulltext(v.clone())))
+        }
    }
 }

--- a/src/common/meta/src/ddl/alter_table/update_metadata.rs
+++ b/src/common/meta/src/ddl/alter_table/update_metadata.rs
@@ -53,7 +53,8 @@ impl AlterTableProcedure {
            }
            AlterKind::DropColumns { .. }
            | AlterKind::ChangeColumnTypes { .. }
-            | AlterKind::ChangeTableOptions { .. } => {}
+            | AlterKind::ChangeTableOptions { .. }
+            | AlterKind::ChangeColumnFulltext { .. } => {}
        }

        Ok(new_info)
--- a/src/common/meta/src/key.rs
+++ b/src/common/meta/src/key.rs
@@ -91,6 +91,7 @@ pub mod catalog_name;
 pub mod datanode_table;
 pub mod flow;
 pub mod node_address;
+mod schema_metadata_manager;
 pub mod schema_name;
 pub mod table_info;
 pub mod table_name;
@@ -116,6 +117,7 @@ use flow::flow_route::FlowRouteValue;
 use flow::table_flow::TableFlowValue;
 use lazy_static::lazy_static;
 use regex::Regex;
+pub use schema_metadata_manager::{SchemaMetadataManager, SchemaMetadataManagerRef};
 use serde::de::DeserializeOwned;
 use serde::{Deserialize, Serialize};
 use snafu::{ensure, OptionExt, ResultExt};
--- a/src/common/meta/src/key/schema_metadata_manager.rs
+++ b/src/common/meta/src/key/schema_metadata_manager.rs
@@ -0,0 +1,122 @@
+// 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.
+
+//! Schema-level metadata manager.
+
+use std::sync::Arc;
+
+use snafu::OptionExt;
+use store_api::storage::TableId;
+
+use crate::error::TableInfoNotFoundSnafu;
+use crate::key::schema_name::{SchemaManager, SchemaNameKey};
+use crate::key::table_info::{TableInfoManager, TableInfoManagerRef};
+use crate::kv_backend::KvBackendRef;
+use crate::{error, SchemaOptions};
+
+pub type SchemaMetadataManagerRef = Arc<SchemaMetadataManager>;
+
+pub struct SchemaMetadataManager {
+    table_info_manager: TableInfoManagerRef,
+    schema_manager: SchemaManager,
+    #[cfg(any(test, feature = "testing"))]
+    kv_backend: KvBackendRef,
+}
+
+impl SchemaMetadataManager {
+    /// Creates a new database meta
+    #[cfg(not(any(test, feature = "testing")))]
+    pub fn new(kv_backend: KvBackendRef) -> Self {
+        let table_info_manager = Arc::new(TableInfoManager::new(kv_backend.clone()));
+        let schema_manager = SchemaManager::new(kv_backend);
+        Self {
+            table_info_manager,
+            schema_manager,
+        }
+    }
+
+    /// Creates a new database meta
+    #[cfg(any(test, feature = "testing"))]
+    pub fn new(kv_backend: KvBackendRef) -> Self {
+        let table_info_manager = Arc::new(TableInfoManager::new(kv_backend.clone()));
+        let schema_manager = SchemaManager::new(kv_backend.clone());
+        Self {
+            table_info_manager,
+            schema_manager,
+            kv_backend,
+        }
+    }
+
+    /// Gets schema options by table id.
+    pub async fn get_schema_options_by_table_id(
+        &self,
+        table_id: TableId,
+    ) -> error::Result<Option<SchemaOptions>> {
+        let table_info = self
+            .table_info_manager
+            .get(table_id)
+            .await?
+            .with_context(|| TableInfoNotFoundSnafu {
+                table: format!("table id: {}", table_id),
+            })?;
+
+        let key = SchemaNameKey::new(
+            &table_info.table_info.catalog_name,
+            &table_info.table_info.schema_name,
+        );
+        self.schema_manager.get(key).await
+    }
+
+    #[cfg(any(test, feature = "testing"))]
+    pub async fn register_region_table_info(
+        &self,
+        table_id: TableId,
+        table_name: &str,
+        schema_name: &str,
+        catalog_name: &str,
+        schema_value: Option<crate::key::schema_name::SchemaNameValue>,
+    ) {
+        use table::metadata::{RawTableInfo, TableType};
+        let value = crate::key::table_info::TableInfoValue::new(RawTableInfo {
+            ident: Default::default(),
+            name: table_name.to_string(),
+            desc: None,
+            catalog_name: catalog_name.to_string(),
+            schema_name: schema_name.to_string(),
+            meta: Default::default(),
+            table_type: TableType::Base,
+        });
+        let (txn, _) = self
+            .table_info_manager
+            .build_create_txn(table_id, &value)
+            .unwrap();
+        let resp = self.kv_backend.txn(txn).await.unwrap();
+        assert!(resp.succeeded, "Failed to create table metadata");
+        let key = SchemaNameKey {
+            catalog: catalog_name,
+            schema: schema_name,
+        };
+        self.schema_manager
+            .create(key, schema_value, false)
+            .await
+            .expect("Failed to create schema metadata");
+        common_telemetry::info!(
+            "Register table: {}, id: {}, schema: {}, catalog: {}",
+            table_name,
+            table_id,
+            schema_name,
+            catalog_name
+        );
+    }
+}
--- a/src/common/meta/src/key/table_info.rs
+++ b/src/common/meta/src/key/table_info.rs
@@ -134,6 +134,7 @@ impl TableInfoValue {
 }

 pub type TableInfoManagerRef = Arc<TableInfoManager>;
+
 #[derive(Clone)]
 pub struct TableInfoManager {
    kv_backend: KvBackendRef,
--- a/src/common/meta/src/lib.rs
+++ b/src/common/meta/src/lib.rs
@@ -54,4 +54,7 @@ pub type DatanodeId = u64;
 // The id of the flownode.
 pub type FlownodeId = u64;

+/// Schema options.
+pub type SchemaOptions = key::schema_name::SchemaNameValue;
+
 pub use instruction::RegionIdent;
--- a/src/common/options/Cargo.toml
+++ b/src/common/options/Cargo.toml
@@ -0,0 +1,13 @@
+[package]
+name = "common-options"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+
+[dependencies]
+common-grpc.workspace = true
+humantime-serde.workspace = true
+serde.workspace = true
+
+[lints]
+workspace = true
--- a/src/frontend/src/service_config/datanode.rs
+++ b/src/frontend/src/service_config/datanode.rs
@@ -18,20 +18,23 @@ use common_grpc::channel_manager;
 use serde::{Deserialize, Serialize};

 #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
-pub struct DatanodeOptions {
-    pub client: DatanodeClientOptions,
+pub struct DatanodeClientOptions {
+    pub client: ClientOptions,
 }

 #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
-pub struct DatanodeClientOptions {
+pub struct ClientOptions {
+    #[serde(with = "humantime_serde")]
+    pub timeout: Duration,
    #[serde(with = "humantime_serde")]
    pub connect_timeout: Duration,
    pub tcp_nodelay: bool,
 }

-impl Default for DatanodeClientOptions {
+impl Default for ClientOptions {
    fn default() -> Self {
        Self {
+            timeout: Duration::from_secs(channel_manager::DEFAULT_GRPC_REQUEST_TIMEOUT_SECS),
            connect_timeout: Duration::from_secs(
                channel_manager::DEFAULT_GRPC_CONNECT_TIMEOUT_SECS,
            ),
--- a/src/servers/src/otlp/plugin.rs
+++ b/src/servers/src/otlp/plugin.rs
@@ -12,17 +12,4 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-use std::sync::Arc;
-
-use opentelemetry_proto::tonic::collector::trace::v1::ExportTraceServiceRequest;
-
-use super::trace::span::TraceSpans;
-
-/// Transformer helps to transform ExportTraceServiceRequest based on logic, like:
-///   - uplift some fields from Attributes (Map type) to column
-pub trait TraceParser: Send + Sync {
-    fn parse(&self, request: ExportTraceServiceRequest) -> TraceSpans;
-    fn table_name(&self) -> String;
-}
-
-pub type TraceParserRef = Arc<dyn TraceParser>;
+pub mod datanode;
--- a/src/datanode/src/alive_keeper.rs
+++ b/src/datanode/src/alive_keeper.rs
@@ -427,7 +427,8 @@ mod test {
        common_telemetry::init_default_ut_logging();
        let mut region_server = mock_region_server();
        let mut engine_env = TestEnv::with_prefix("region-alive-keeper");
-        let engine = Arc::new(engine_env.create_engine(MitoConfig::default()).await);
+        let engine = engine_env.create_engine(MitoConfig::default()).await;
+        let engine = Arc::new(engine);
        region_server.register_engine(engine.clone());

        let alive_keeper = Arc::new(RegionAliveKeeper::new(region_server.clone(), 100));
--- a/src/datanode/src/config.rs
+++ b/src/datanode/src/config.rs
@@ -30,7 +30,7 @@ use servers::heartbeat_options::HeartbeatOptions;
 use servers::http::HttpOptions;
 use servers::Mode;

-pub const DEFAULT_OBJECT_STORE_CACHE_SIZE: ReadableSize = ReadableSize::mb(256);
+pub const DEFAULT_OBJECT_STORE_CACHE_SIZE: ReadableSize = ReadableSize::gb(1);

 /// Default data home in file storage
 const DEFAULT_DATA_HOME: &str = "/tmp/greptimedb";
--- a/src/datanode/src/datanode.rs
+++ b/src/datanode/src/datanode.rs
@@ -23,6 +23,7 @@ use common_base::Plugins;
 use common_error::ext::BoxedError;
 use common_greptimedb_telemetry::GreptimeDBTelemetryTask;
 use common_meta::key::datanode_table::{DatanodeTableManager, DatanodeTableValue};
+use common_meta::key::{SchemaMetadataManager, SchemaMetadataManagerRef};
 use common_meta::kv_backend::KvBackendRef;
 use common_meta::wal_options_allocator::prepare_wal_options;
 pub use common_procedure::options::ProcedureConfig;
@@ -207,7 +208,10 @@ impl DatanodeBuilder {
            (Box::new(NoopRegionServerEventListener) as _, None)
        };

-        let region_server = self.new_region_server(region_event_listener).await?;
+        let schema_metadata_manager = Arc::new(SchemaMetadataManager::new(kv_backend.clone()));
+        let region_server = self
+            .new_region_server(schema_metadata_manager, region_event_listener)
+            .await?;

        let datanode_table_manager = DatanodeTableManager::new(kv_backend.clone());
        let table_values = datanode_table_manager
@@ -312,6 +316,7 @@ impl DatanodeBuilder {

    async fn new_region_server(
        &self,
+        schema_metadata_manager: SchemaMetadataManagerRef,
        event_listener: RegionServerEventListenerRef,
    ) -> Result<RegionServer> {
        let opts: &DatanodeOptions = &self.opts;
@@ -340,8 +345,13 @@ impl DatanodeBuilder {
        );

        let object_store_manager = Self::build_object_store_manager(&opts.storage).await?;
-        let engines =
-            Self::build_store_engines(opts, object_store_manager, self.plugins.clone()).await?;
+        let engines = Self::build_store_engines(
+            opts,
+            object_store_manager,
+            schema_metadata_manager,
+            self.plugins.clone(),
+        )
+        .await?;
        for engine in engines {
            region_server.register_engine(engine);
        }
@@ -355,6 +365,7 @@ impl DatanodeBuilder {
    async fn build_store_engines(
        opts: &DatanodeOptions,
        object_store_manager: ObjectStoreManagerRef,
+        schema_metadata_manager: SchemaMetadataManagerRef,
        plugins: Plugins,
    ) -> Result<Vec<RegionEngineRef>> {
        let mut engines = vec![];
@@ -365,6 +376,7 @@ impl DatanodeBuilder {
                        opts,
                        object_store_manager.clone(),
                        config.clone(),
+                        schema_metadata_manager.clone(),
                        plugins.clone(),
                    )
                    .await?;
@@ -390,6 +402,7 @@ impl DatanodeBuilder {
        opts: &DatanodeOptions,
        object_store_manager: ObjectStoreManagerRef,
        config: MitoConfig,
+        schema_metadata_manager: SchemaMetadataManagerRef,
        plugins: Plugins,
    ) -> Result<MitoEngine> {
        let mito_engine = match &opts.wal {
@@ -399,6 +412,7 @@ impl DatanodeBuilder {
                Self::build_raft_engine_log_store(&opts.storage.data_home, raft_engine_config)
                    .await?,
                object_store_manager,
+                schema_metadata_manager,
                plugins,
            )
            .await
@@ -429,6 +443,7 @@ impl DatanodeBuilder {
                    config,
                    Self::build_kafka_log_store(kafka_config, global_index_collector).await?,
                    object_store_manager,
+                    schema_metadata_manager,
                    plugins,
                )
                .await
--- a/src/datanode/src/region_server.rs
+++ b/src/datanode/src/region_server.rs
@@ -1355,7 +1355,7 @@ mod tests {
    }

    #[tokio::test]
-    async fn test_region_server_parallism() {
+    async fn test_region_server_parallelism() {
        let p = RegionServerParallelism::from_opts(2, Duration::from_millis(1)).unwrap();
        let first_query = p.acquire().await;
        assert!(first_query.is_ok());
--- a/src/datatypes/Cargo.toml
+++ b/src/datatypes/Cargo.toml
@@ -33,3 +33,5 @@ paste = "1.0"
 serde.workspace = true
 serde_json.workspace = true
 snafu.workspace = true
+sqlparser.workspace = true
+sqlparser_derive = "0.1"
--- a/src/datatypes/src/data_type.rs
+++ b/src/datatypes/src/data_type.rs
@@ -36,7 +36,7 @@ use crate::types::{
    IntervalDayTimeType, IntervalMonthDayNanoType, IntervalType, IntervalYearMonthType, JsonType,
    ListType, NullType, StringType, TimeMillisecondType, TimeType, TimestampMicrosecondType,
    TimestampMillisecondType, TimestampNanosecondType, TimestampSecondType, TimestampType,
-    UInt16Type, UInt32Type, UInt64Type, UInt8Type,
+    UInt16Type, UInt32Type, UInt64Type, UInt8Type, VectorType,
 };
 use crate::value::Value;
 use crate::vectors::MutableVector;
@@ -84,6 +84,9 @@ pub enum ConcreteDataType {

    // JSON type:
    Json(JsonType),
+
+    // Vector type:
+    Vector(VectorType),
 }

 impl fmt::Display for ConcreteDataType {
@@ -132,6 +135,7 @@ impl fmt::Display for ConcreteDataType {
            ConcreteDataType::List(v) => write!(f, "{}", v.name()),
            ConcreteDataType::Dictionary(v) => write!(f, "{}", v.name()),
            ConcreteDataType::Json(v) => write!(f, "{}", v.name()),
+            ConcreteDataType::Vector(v) => write!(f, "{}", v.name()),
        }
    }
 }
@@ -167,6 +171,7 @@ impl ConcreteDataType {
                | ConcreteDataType::Decimal128(_)
                | ConcreteDataType::Binary(_)
                | ConcreteDataType::Json(_)
+                | ConcreteDataType::Vector(_)
        )
    }

@@ -225,6 +230,10 @@ impl ConcreteDataType {
        matches!(self, ConcreteDataType::Json(_))
    }

+    pub fn is_vector(&self) -> bool {
+        matches!(self, ConcreteDataType::Vector(_))
+    }
+
    pub fn numerics() -> Vec<ConcreteDataType> {
        vec![
            ConcreteDataType::int8_datatype(),
@@ -334,6 +343,20 @@ impl ConcreteDataType {
        }
    }

+    pub fn as_json(&self) -> Option<JsonType> {
+        match self {
+            ConcreteDataType::Json(j) => Some(*j),
+            _ => None,
+        }
+    }
+
+    pub fn as_vector(&self) -> Option<VectorType> {
+        match self {
+            ConcreteDataType::Vector(v) => Some(*v),
+            _ => None,
+        }
+    }
+
    /// Checks if the data type can cast to another data type.
    pub fn can_arrow_type_cast_to(&self, to_type: &ConcreteDataType) -> bool {
        let array = arrow_array::new_empty_array(&self.as_arrow_type());
@@ -564,6 +587,14 @@ impl ConcreteDataType {
    pub fn decimal128_default_datatype() -> ConcreteDataType {
        Self::decimal128_datatype(DECIMAL128_MAX_PRECISION, DECIMAL_DEFAULT_SCALE)
    }
+
+    pub fn vector_datatype(dim: u32) -> ConcreteDataType {
+        ConcreteDataType::Vector(VectorType::new(dim))
+    }
+
+    pub fn vector_default_datatype() -> ConcreteDataType {
+        Self::vector_datatype(0)
+    }
 }

 /// Data type abstraction.
@@ -757,6 +788,7 @@ mod tests {
        assert!(ConcreteDataType::duration_microsecond_datatype().is_stringifiable());
        assert!(ConcreteDataType::duration_nanosecond_datatype().is_stringifiable());
        assert!(ConcreteDataType::decimal128_datatype(10, 2).is_stringifiable());
+        assert!(ConcreteDataType::vector_default_datatype().is_stringifiable());
    }

    #[test]
@@ -909,5 +941,9 @@ mod tests {
            .to_string(),
            "Dictionary<Int32, String>"
        );
+        assert_eq!(
+            ConcreteDataType::vector_datatype(3).to_string(),
+            "Vector(3)"
+        );
    }
 }
--- a/src/datatypes/src/error.rs
+++ b/src/datatypes/src/error.rs
@@ -196,6 +196,13 @@ pub enum Error {
        location: Location,
    },

+    #[snafu(display("Invalid Vector: {}", msg))]
+    InvalidVector {
+        msg: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
    #[snafu(display("Value exceeds the precision {} bound", precision))]
    ValueExceedsPrecision {
        precision: u8,
@@ -212,6 +219,19 @@ pub enum Error {
        #[snafu(implicit)]
        location: Location,
    },
+
+    #[snafu(display("Failed to parse extended type in metadata: {}", value))]
+    ParseExtendedType {
+        value: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
+    #[snafu(display("Invalid fulltext option: {}", msg))]
+    InvalidFulltextOption {
+        msg: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
 }

 impl ErrorExt for Error {
@@ -230,7 +250,9 @@ impl ErrorExt for Error {
            | DuplicateMeta { .. }
            | InvalidTimestampPrecision { .. }
            | InvalidPrecisionOrScale { .. }
-            | InvalidJson { .. } => StatusCode::InvalidArguments,
+            | InvalidJson { .. }
+            | InvalidVector { .. }
+            | InvalidFulltextOption { .. } => StatusCode::InvalidArguments,

            ValueExceedsPrecision { .. }
            | CastType { .. }
@@ -245,7 +267,8 @@ impl ErrorExt for Error {
            | ProjectArrowSchema { .. }
            | ToScalarValue { .. }
            | TryFromValue { .. }
-            | ConvertArrowArrayToScalars { .. } => StatusCode::Internal,
+            | ConvertArrowArrayToScalars { .. }
+            | ParseExtendedType { .. } => StatusCode::Internal,
        }
    }

--- a/src/datatypes/src/schema.rs
+++ b/src/datatypes/src/schema.rs
@@ -21,13 +21,16 @@ use std::fmt;
 use std::sync::Arc;

 use arrow::datatypes::{Field, Schema as ArrowSchema};
+use column_schema::ColumnExtType;
 use datafusion_common::DFSchemaRef;
 use snafu::{ensure, ResultExt};

 use crate::error::{self, DuplicateColumnSnafu, Error, ProjectArrowSchemaSnafu, Result};
-use crate::prelude::DataType;
+use crate::prelude::ConcreteDataType;
 pub use crate::schema::column_schema::{
-    ColumnSchema, FulltextAnalyzer, FulltextOptions, Metadata, COMMENT_KEY, FULLTEXT_KEY,
+    ColumnSchema, FulltextAnalyzer, FulltextOptions, Metadata,
+    COLUMN_FULLTEXT_CHANGE_OPT_KEY_ENABLE, COLUMN_FULLTEXT_OPT_KEY_ANALYZER,
+    COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE, COMMENT_KEY, FULLTEXT_KEY, INVERTED_INDEX_KEY,
    TIME_INDEX_KEY,
 };
 pub use crate::schema::constraint::ColumnDefaultConstraint;
@@ -261,9 +264,14 @@ fn collect_fields(column_schemas: &[ColumnSchema]) -> Result<FieldsAndIndices> {
        }
        let mut field = Field::try_from(column_schema)?;

-        // Json column performs the same as binary column in Arrow, so we need to mark it
-        if column_schema.data_type.is_json() {
-            let metadata = HashMap::from([(TYPE_KEY.to_string(), column_schema.data_type.name())]);
+        // Column with type Json or Vector performs the same as binary column in Arrow, so we need to mark it
+        let extype = match column_schema.data_type {
+            ConcreteDataType::Json(_) => Some(ColumnExtType::Json),
+            ConcreteDataType::Vector(d) => Some(ColumnExtType::Vector(d.dim)),
+            _ => None,
+        };
+        if let Some(extype) = extype {
+            let metadata = HashMap::from([(TYPE_KEY.to_string(), extype.to_string())]);
            field = field.with_metadata(metadata);
        }
        fields.push(field);
--- a/src/datatypes/src/schema/column_schema.rs
+++ b/src/datatypes/src/schema/column_schema.rs
@@ -14,16 +14,17 @@

 use std::collections::HashMap;
 use std::fmt;
+use std::str::FromStr;

 use arrow::datatypes::Field;
 use serde::{Deserialize, Serialize};
 use snafu::{ensure, ResultExt};
+use sqlparser_derive::{Visit, VisitMut};

 use crate::data_type::{ConcreteDataType, DataType};
-use crate::error::{self, Error, Result};
+use crate::error::{self, Error, InvalidFulltextOptionSnafu, ParseExtendedTypeSnafu, Result};
 use crate::schema::constraint::ColumnDefaultConstraint;
 use crate::schema::TYPE_KEY;
-use crate::types::JSON_TYPE_NAME;
 use crate::value::Value;
 use crate::vectors::VectorRef;

@@ -36,6 +37,13 @@ pub const COMMENT_KEY: &str = "greptime:storage:comment";
 const DEFAULT_CONSTRAINT_KEY: &str = "greptime:default_constraint";
 /// Key used to store fulltext options in arrow field's metadata.
 pub const FULLTEXT_KEY: &str = "greptime:fulltext";
+/// Key used to store whether the column has inverted index in arrow field's metadata.
+pub const INVERTED_INDEX_KEY: &str = "greptime:inverted_index";
+
+/// Keys used in fulltext options
+pub const COLUMN_FULLTEXT_CHANGE_OPT_KEY_ENABLE: &str = "enable";
+pub const COLUMN_FULLTEXT_OPT_KEY_ANALYZER: &str = "analyzer";
+pub const COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE: &str = "case_sensitive";

 /// Schema of a column, used as an immutable struct.
 #[derive(Clone, PartialEq, Eq, Serialize, Deserialize)]
@@ -134,6 +142,24 @@ impl ColumnSchema {
        self
    }

+    pub fn set_inverted_index(mut self, value: bool) -> Self {
+        let _ = self
+            .metadata
+            .insert(INVERTED_INDEX_KEY.to_string(), value.to_string());
+        self
+    }
+
+    pub fn is_inverted_indexed(&self) -> bool {
+        self.metadata
+            .get(INVERTED_INDEX_KEY)
+            .map(|v| v.eq_ignore_ascii_case("true"))
+            .unwrap_or(false)
+    }
+
+    pub fn has_inverted_index_key(&self) -> bool {
+        self.metadata.contains_key(INVERTED_INDEX_KEY)
+    }
+
    /// Set default constraint.
    ///
    /// If a default constraint exists for the column, this method will
@@ -264,6 +290,48 @@ impl ColumnSchema {
        );
        Ok(self)
    }
+
+    pub fn set_fulltext_options(&mut self, options: &FulltextOptions) -> Result<()> {
+        self.metadata.insert(
+            FULLTEXT_KEY.to_string(),
+            serde_json::to_string(options).context(error::SerializeSnafu)?,
+        );
+        Ok(())
+    }
+}
+
+/// Column extended type set in column schema's metadata.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum ColumnExtType {
+    /// Json type.
+    Json,
+
+    /// Vector type with dimension.
+    Vector(u32),
+}
+
+impl fmt::Display for ColumnExtType {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        match self {
+            ColumnExtType::Json => write!(f, "Json"),
+            ColumnExtType::Vector(dim) => write!(f, "Vector({})", dim),
+        }
+    }
+}
+
+impl FromStr for ColumnExtType {
+    type Err = String;
+
+    fn from_str(s: &str) -> std::result::Result<Self, Self::Err> {
+        match s {
+            "Json" => Ok(ColumnExtType::Json),
+            _ if s.starts_with("Vector(") && s.ends_with(')') => s[7..s.len() - 1]
+                .parse::<u32>()
+                .map(ColumnExtType::Vector)
+                .map_err(|_| "Invalid dimension for Vector".to_string()),
+            _ => Err("Unknown variant".to_string()),
+        }
+    }
 }

 impl TryFrom<&Field> for ColumnSchema {
@@ -272,11 +340,17 @@ impl TryFrom<&Field> for ColumnSchema {
    fn try_from(field: &Field) -> Result<ColumnSchema> {
        let mut data_type = ConcreteDataType::try_from(field.data_type())?;
        // Override the data type if it is specified in the metadata.
-        if field.metadata().contains_key(TYPE_KEY) {
-            data_type = match field.metadata().get(TYPE_KEY).unwrap().as_str() {
-                JSON_TYPE_NAME => ConcreteDataType::json_datatype(),
-                _ => data_type,
-            };
+        if let Some(s) = field.metadata().get(TYPE_KEY) {
+            let extype = ColumnExtType::from_str(s)
+                .map_err(|_| ParseExtendedTypeSnafu { value: s }.build())?;
+            match extype {
+                ColumnExtType::Json => {
+                    data_type = ConcreteDataType::json_datatype();
+                }
+                ColumnExtType::Vector(dim) => {
+                    data_type = ConcreteDataType::vector_datatype(dim);
+                }
+            }
        }
        let mut metadata = field.metadata().clone();
        let default_constraint = match metadata.remove(DEFAULT_CONSTRAINT_KEY) {
@@ -328,7 +402,7 @@ impl TryFrom<&ColumnSchema> for Field {
 }

 /// Fulltext options for a column.
-#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default, Visit, VisitMut)]
 #[serde(rename_all = "kebab-case")]
 pub struct FulltextOptions {
    /// Whether the fulltext index is enabled.
@@ -341,8 +415,71 @@ pub struct FulltextOptions {
    pub case_sensitive: bool,
 }

+impl fmt::Display for FulltextOptions {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(f, "enable={}", self.enable)?;
+        if self.enable {
+            write!(f, ", analyzer={}", self.analyzer)?;
+            write!(f, ", case_sensitive={}", self.case_sensitive)?;
+        }
+        Ok(())
+    }
+}
+
+impl TryFrom<HashMap<String, String>> for FulltextOptions {
+    type Error = Error;
+
+    fn try_from(options: HashMap<String, String>) -> Result<Self> {
+        let mut fulltext_options = FulltextOptions {
+            enable: true,
+            ..Default::default()
+        };
+
+        if let Some(enable) = options.get(COLUMN_FULLTEXT_CHANGE_OPT_KEY_ENABLE) {
+            match enable.to_ascii_lowercase().as_str() {
+                "true" => fulltext_options.enable = true,
+                "false" => fulltext_options.enable = false,
+                _ => {
+                    return InvalidFulltextOptionSnafu {
+                        msg: format!("{enable}, expected: 'true' | 'false'"),
+                    }
+                    .fail();
+                }
+            }
+        };
+
+        if let Some(analyzer) = options.get(COLUMN_FULLTEXT_OPT_KEY_ANALYZER) {
+            match analyzer.to_ascii_lowercase().as_str() {
+                "english" => fulltext_options.analyzer = FulltextAnalyzer::English,
+                "chinese" => fulltext_options.analyzer = FulltextAnalyzer::Chinese,
+                _ => {
+                    return InvalidFulltextOptionSnafu {
+                        msg: format!("{analyzer}, expected: 'English' | 'Chinese'"),
+                    }
+                    .fail();
+                }
+            }
+        };
+
+        if let Some(case_sensitive) = options.get(COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE) {
+            match case_sensitive.to_ascii_lowercase().as_str() {
+                "true" => fulltext_options.case_sensitive = true,
+                "false" => fulltext_options.case_sensitive = false,
+                _ => {
+                    return InvalidFulltextOptionSnafu {
+                        msg: format!("{case_sensitive}, expected: 'true' | 'false'"),
+                    }
+                    .fail();
+                }
+            }
+        }
+
+        Ok(fulltext_options)
+    }
+}
+
 /// Fulltext analyzer.
-#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default, Visit, VisitMut)]
 pub enum FulltextAnalyzer {
    #[default]
    English,
@@ -564,5 +701,24 @@ mod tests {
            column_schema.metadata.get(TYPE_KEY).unwrap(),
            &ConcreteDataType::json_datatype().name()
        );
+
+        let field = Field::new("test", ArrowDataType::Binary, true);
+        let field = field.with_metadata(Metadata::from([(
+            TYPE_KEY.to_string(),
+            ConcreteDataType::vector_datatype(3).name(),
+        )]));
+        let column_schema = ColumnSchema::try_from(&field).unwrap();
+        assert_eq!("test", column_schema.name);
+        assert_eq!(
+            ConcreteDataType::vector_datatype(3),
+            column_schema.data_type
+        );
+        assert!(column_schema.is_nullable);
+        assert!(!column_schema.is_time_index);
+        assert!(column_schema.default_constraint.is_none());
+        assert_eq!(
+            column_schema.metadata.get(TYPE_KEY).unwrap(),
+            &ConcreteDataType::vector_datatype(3).name()
+        );
    }
 }
--- a/src/datatypes/src/schema/constraint.rs
+++ b/src/datatypes/src/schema/constraint.rs
@@ -94,7 +94,7 @@ impl ColumnDefaultConstraint {
                    // Whether the value could be nullable has been checked before, only need
                    // to check the type compatibility here.
                    ensure!(
-                        data_type.logical_type_id() == v.logical_type_id(),
+                        value_type_match(data_type, v.data_type()),
                        error::DefaultValueTypeSnafu {
                            reason: format!(
                                "column has type {:?} but default value has type {:?}",
@@ -215,6 +215,17 @@ fn create_current_timestamp_vector(
    }
 }

+fn value_type_match(column_type: &ConcreteDataType, value_type: ConcreteDataType) -> bool {
+    match (column_type, value_type) {
+        (ct, vt) if ct.logical_type_id() == vt.logical_type_id() => true,
+        // Vector and Json type is encoded as binary
+        (ConcreteDataType::Vector(_) | ConcreteDataType::Json(_), ConcreteDataType::Binary(_)) => {
+            true
+        }
+        _ => false,
+    }
+}
+
 #[cfg(test)]
 mod tests {
    use std::sync::Arc;
--- a/src/datatypes/src/type_id.rs
+++ b/src/datatypes/src/type_id.rs
@@ -70,6 +70,8 @@ pub enum LogicalTypeId {
    Dictionary,

    Json,
+
+    Vector,
 }

 impl LogicalTypeId {
@@ -129,6 +131,7 @@ impl LogicalTypeId {
            LogicalTypeId::DurationNanosecond => ConcreteDataType::duration_nanosecond_datatype(),
            LogicalTypeId::Decimal128 => ConcreteDataType::decimal128_default_datatype(),
            LogicalTypeId::Json => ConcreteDataType::json_datatype(),
+            LogicalTypeId::Vector => ConcreteDataType::vector_default_datatype(),
        }
    }
 }
--- a/src/datatypes/src/types.rs
+++ b/src/datatypes/src/types.rs
@@ -28,6 +28,7 @@ mod primitive_type;
 mod string_type;
 mod time_type;
 mod timestamp_type;
+mod vector_type;

 pub use binary_type::BinaryType;
 pub use boolean_type::BooleanType;
@@ -43,7 +44,9 @@ pub use duration_type::{
 pub use interval_type::{
    IntervalDayTimeType, IntervalMonthDayNanoType, IntervalType, IntervalYearMonthType,
 };
-pub use json_type::{JsonType, JSON_TYPE_NAME};
+pub use json_type::{
+    json_type_value_to_string, parse_string_to_json_type_value, JsonType, JSON_TYPE_NAME,
+};
 pub use list_type::ListType;
 pub use null_type::NullType;
 pub use primitive_type::{
@@ -58,3 +61,4 @@ pub use timestamp_type::{
    TimestampMicrosecondType, TimestampMillisecondType, TimestampNanosecondType,
    TimestampSecondType, TimestampType,
 };
+pub use vector_type::{parse_string_to_vector_type_value, vector_type_value_to_string, VectorType};
--- a/src/datatypes/src/types/json_type.rs
+++ b/src/datatypes/src/types/json_type.rs
@@ -12,13 +12,12 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-use std::sync::Arc;
-
 use arrow::datatypes::DataType as ArrowDataType;
 use common_base::bytes::Bytes;
 use serde::{Deserialize, Serialize};

-use crate::data_type::{DataType, DataTypeRef};
+use crate::data_type::DataType;
+use crate::error::{InvalidJsonSnafu, Result};
 use crate::scalars::ScalarVectorBuilder;
 use crate::type_id::LogicalTypeId;
 use crate::value::Value;
@@ -26,14 +25,29 @@ use crate::vectors::{BinaryVectorBuilder, MutableVector};

 pub const JSON_TYPE_NAME: &str = "Json";

+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+pub enum JsonFormat {
+    Jsonb,
+}
+
+impl Default for JsonFormat {
+    fn default() -> Self {
+        Self::Jsonb
+    }
+}
+
 /// JsonType is a data type for JSON data. It is stored as binary data of jsonb format.
 /// It utilizes current binary value and vector implementation.
-#[derive(Debug, Default, Clone, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
-pub struct JsonType;
+#[derive(
+    Debug, Default, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize,
+)]
+pub struct JsonType {
+    pub format: JsonFormat,
+}

 impl JsonType {
-    pub fn arc() -> DataTypeRef {
-        Arc::new(Self)
+    pub fn new(format: JsonFormat) -> Self {
+        Self { format }
    }
 }

@@ -65,3 +79,19 @@ impl DataType for JsonType {
        }
    }
 }
+
+/// Converts a json type value to string
+pub fn json_type_value_to_string(val: &[u8], format: &JsonFormat) -> Result<String> {
+    match format {
+        JsonFormat::Jsonb => Ok(jsonb::to_string(val)),
+    }
+}
+
+/// Parses a string to a json type value
+pub fn parse_string_to_json_type_value(s: &str, format: &JsonFormat) -> Result<Vec<u8>> {
+    match format {
+        JsonFormat::Jsonb => jsonb::parse_value(s.as_bytes())
+            .map_err(|_| InvalidJsonSnafu { value: s }.build())
+            .map(|json| json.to_vec()),
+    }
+}
--- a/src/datatypes/src/types/vector_type.rs
+++ b/src/datatypes/src/types/vector_type.rs
@@ -0,0 +1,237 @@
+// 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;
+use common_base::bytes::Bytes;
+use serde::{Deserialize, Serialize};
+
+use crate::data_type::DataType;
+use crate::error::{InvalidVectorSnafu, Result};
+use crate::scalars::ScalarVectorBuilder;
+use crate::type_id::LogicalTypeId;
+use crate::value::Value;
+use crate::vectors::{BinaryVectorBuilder, MutableVector};
+
+/// `VectorType` is a data type for vector data with a fixed dimension.
+/// The type of items in the vector is float32.
+/// It is stored as binary data that contains the concatenated float32 values.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
+pub struct VectorType {
+    pub dim: u32,
+}
+
+impl VectorType {
+    pub fn new(dim: u32) -> Self {
+        Self { dim }
+    }
+}
+
+impl DataType for VectorType {
+    fn name(&self) -> String {
+        format!("Vector({})", self.dim)
+    }
+
+    fn logical_type_id(&self) -> LogicalTypeId {
+        LogicalTypeId::Vector
+    }
+
+    fn default_value(&self) -> Value {
+        Bytes::default().into()
+    }
+
+    fn as_arrow_type(&self) -> ArrowDataType {
+        ArrowDataType::Binary
+    }
+
+    fn create_mutable_vector(&self, capacity: usize) -> Box<dyn MutableVector> {
+        Box::new(BinaryVectorBuilder::with_capacity(capacity))
+    }
+
+    fn try_cast(&self, from: Value) -> Option<Value> {
+        match from {
+            Value::Binary(v) => Some(Value::Binary(v)),
+            _ => None,
+        }
+    }
+}
+
+/// Converts a vector type value to string
+/// for example: [1.0, 2.0, 3.0] -> "[1.0,2.0,3.0]"
+pub fn vector_type_value_to_string(val: &[u8], dim: u32) -> Result<String> {
+    let expected_len = dim as usize * std::mem::size_of::<f32>();
+    if val.len() != expected_len {
+        return InvalidVectorSnafu {
+            msg: format!(
+                "Failed to convert Vector value to string: wrong byte size, expected {}, got {}",
+                expected_len,
+                val.len()
+            ),
+        }
+        .fail();
+    }
+
+    if dim == 0 {
+        return Ok("[]".to_string());
+    }
+
+    let elements = unsafe {
+        std::slice::from_raw_parts(
+            val.as_ptr() as *const f32,
+            val.len() / std::mem::size_of::<f32>(),
+        )
+    };
+
+    let mut s = String::from("[");
+    for (i, e) in elements.iter().enumerate() {
+        if i > 0 {
+            s.push(',');
+        }
+        s.push_str(&e.to_string());
+    }
+    s.push(']');
+    Ok(s)
+}
+
+/// Parses a string to a vector type value
+/// Valid input format: "[1.0,2.0,3.0]", "[1.0, 2.0, 3.0]"
+pub fn parse_string_to_vector_type_value(s: &str, dim: u32) -> Result<Vec<u8>> {
+    // Trim the brackets
+    let trimmed = s.trim();
+    if !trimmed.starts_with('[') || !trimmed.ends_with(']') {
+        return InvalidVectorSnafu {
+            msg: format!("Failed to parse {s} to Vector value: not properly enclosed in brackets"),
+        }
+        .fail();
+    }
+    // Remove the brackets
+    let content = trimmed[1..trimmed.len() - 1].trim();
+
+    if content.is_empty() {
+        if dim != 0 {
+            return InvalidVectorSnafu {
+                msg: format!("Failed to parse {s} to Vector value: wrong dimension"),
+            }
+            .fail();
+        }
+        return Ok(vec![]);
+    }
+
+    let elements = content
+        .split(',')
+        .map(|e| {
+            e.trim().parse::<f32>().map_err(|_| {
+                InvalidVectorSnafu {
+                    msg: format!(
+                        "Failed to parse {s} to Vector value: elements are not all float32"
+                    ),
+                }
+                .build()
+            })
+        })
+        .collect::<Result<Vec<f32>>>()?;
+
+    // Check dimension
+    if elements.len() != dim as usize {
+        return InvalidVectorSnafu {
+            msg: format!("Failed to parse {s} to Vector value: wrong dimension"),
+        }
+        .fail();
+    }
+
+    // Convert Vec<f32> to Vec<u8>
+    let bytes = unsafe {
+        std::slice::from_raw_parts(
+            elements.as_ptr() as *const u8,
+            elements.len() * std::mem::size_of::<f32>(),
+        )
+        .to_vec()
+    };
+
+    Ok(bytes)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_conversion_between_string_and_vector_type_value() {
+        let dim = 3;
+
+        let cases = [
+            ("[1.0,2.0,3]", "[1,2,3]"),
+            ("[0.0 , 0.0 , 0.0]", "[0,0,0]"),
+            ("[3.4028235e38, -3.4028235e38, 1.1754944e-38]", "[340282350000000000000000000000000000000,-340282350000000000000000000000000000000,0.000000000000000000000000000000000000011754944]"),
+        ];
+
+        for (s, expected) in cases.iter() {
+            let val = parse_string_to_vector_type_value(s, dim).unwrap();
+            let s = vector_type_value_to_string(&val, dim).unwrap();
+            assert_eq!(s, *expected);
+        }
+
+        let dim = 0;
+        let cases = [("[]", "[]"), ("[ ]", "[]"), ("[  ]", "[]")];
+        for (s, expected) in cases.iter() {
+            let val = parse_string_to_vector_type_value(s, dim).unwrap();
+            let s = vector_type_value_to_string(&val, dim).unwrap();
+            assert_eq!(s, *expected);
+        }
+    }
+
+    #[test]
+    fn test_vector_type_value_to_string_wrong_byte_size() {
+        let dim = 3;
+        let val = vec![1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11];
+        let res = vector_type_value_to_string(&val, dim);
+        assert!(res.is_err());
+
+        let dim = 0;
+        let val = vec![1];
+        let res = vector_type_value_to_string(&val, dim);
+        assert!(res.is_err());
+    }
+
+    #[test]
+    fn test_parse_string_to_vector_type_value_not_properly_enclosed_in_brackets() {
+        let dim = 3;
+        let s = "1.0,2.0,3.0";
+        let res = parse_string_to_vector_type_value(s, dim);
+        assert!(res.is_err());
+
+        let s = "[1.0,2.0,3.0";
+        let res = parse_string_to_vector_type_value(s, dim);
+        assert!(res.is_err());
+
+        let s = "1.0,2.0,3.0]";
+        let res = parse_string_to_vector_type_value(s, dim);
+        assert!(res.is_err());
+    }
+
+    #[test]
+    fn test_parse_string_to_vector_type_value_wrong_dimension() {
+        let dim = 3;
+        let s = "[1.0,2.0]";
+        let res = parse_string_to_vector_type_value(s, dim);
+        assert!(res.is_err());
+    }
+
+    #[test]
+    fn test_parse_string_to_vector_type_value_elements_are_not_all_float32() {
+        let dim = 3;
+        let s = "[1.0,2.0,ah]";
+        let res = parse_string_to_vector_type_value(s, dim);
+        assert!(res.is_err());
+    }
+}
--- a/src/datatypes/src/value.rs
+++ b/src/datatypes/src/value.rs
@@ -520,7 +520,9 @@ pub fn to_null_scalar_value(output_type: &ConcreteDataType) -> Result<ScalarValu
        ConcreteDataType::UInt64(_) => ScalarValue::UInt64(None),
        ConcreteDataType::Float32(_) => ScalarValue::Float32(None),
        ConcreteDataType::Float64(_) => ScalarValue::Float64(None),
-        ConcreteDataType::Binary(_) | ConcreteDataType::Json(_) => ScalarValue::Binary(None),
+        ConcreteDataType::Binary(_) | ConcreteDataType::Json(_) | ConcreteDataType::Vector(_) => {
+            ScalarValue::Binary(None)
+        }
        ConcreteDataType::String(_) => ScalarValue::Utf8(None),
        ConcreteDataType::Date(_) => ScalarValue::Date32(None),
        ConcreteDataType::DateTime(_) => ScalarValue::Date64(None),
--- a/src/datatypes/src/vectors/binary.rs
+++ b/src/datatypes/src/vectors/binary.rs
@@ -20,9 +20,10 @@ use snafu::ResultExt;

 use crate::arrow_array::{BinaryArray, MutableBinaryArray};
 use crate::data_type::ConcreteDataType;
-use crate::error::{self, Result};
+use crate::error::{self, InvalidVectorSnafu, Result};
 use crate::scalars::{ScalarVector, ScalarVectorBuilder};
 use crate::serialize::Serializable;
+use crate::types::parse_string_to_vector_type_value;
 use crate::value::{Value, ValueRef};
 use crate::vectors::{self, MutableVector, Validity, Vector, VectorRef};

@@ -66,6 +67,40 @@ impl BinaryVector {
        }
        Ok(BinaryVector::from(vector))
    }
+
+    pub fn convert_binary_to_vector(&self, dim: u32) -> Result<BinaryVector> {
+        let arrow_array = self.to_arrow_array();
+        let mut vector = vec![];
+        for binary in arrow_array
+            .as_any()
+            .downcast_ref::<BinaryArray>()
+            .unwrap()
+            .iter()
+        {
+            let v = if let Some(binary) = binary {
+                let bytes_size = dim as usize * std::mem::size_of::<f32>();
+                if let Ok(s) = String::from_utf8(binary.to_vec()) {
+                    let v = parse_string_to_vector_type_value(&s, dim)?;
+                    Some(v)
+                } else if binary.len() == dim as usize * std::mem::size_of::<f32>() {
+                    Some(binary.to_vec())
+                } else {
+                    return InvalidVectorSnafu {
+                        msg: format!(
+                            "Unexpected bytes size for vector value, expected {}, got {}",
+                            bytes_size,
+                            binary.len()
+                        ),
+                    }
+                    .fail();
+                }
+            } else {
+                None
+            };
+            vector.push(v);
+        }
+        Ok(BinaryVector::from(vector))
+    }
 }

 impl From<BinaryArray> for BinaryVector {
@@ -462,5 +497,56 @@ mod tests {
            .convert_binary_to_json()
            .unwrap_err();
        assert_matches!(error, error::Error::InvalidJson { .. });
+
+        // corrupted jsonb
+        let jsonb = jsonb::parse_value("{\"hello\": \"world\"}".as_bytes())
+            .unwrap()
+            .to_vec();
+        let corrupted_jsonb = jsonb[0..jsonb.len() - 1].to_vec();
+        let error = BinaryVector::from(vec![corrupted_jsonb])
+            .convert_binary_to_json()
+            .unwrap_err();
+        assert_matches!(error, error::Error::InvalidJson { .. });
+    }
+
+    #[test]
+    fn test_binary_vector_conversion() {
+        let dim = 3;
+        let vector = BinaryVector::from(vec![
+            Some(b"[1,2,3]".to_vec()),
+            Some(b"[4,5,6]".to_vec()),
+            Some(b"[7,8,9]".to_vec()),
+            None,
+        ]);
+        let expected = BinaryVector::from(vec![
+            Some(
+                [1.0f32, 2.0, 3.0]
+                    .iter()
+                    .flat_map(|v| v.to_le_bytes())
+                    .collect(),
+            ),
+            Some(
+                [4.0f32, 5.0, 6.0]
+                    .iter()
+                    .flat_map(|v| v.to_le_bytes())
+                    .collect(),
+            ),
+            Some(
+                [7.0f32, 8.0, 9.0]
+                    .iter()
+                    .flat_map(|v| v.to_le_bytes())
+                    .collect(),
+            ),
+            None,
+        ]);
+
+        let converted = vector.convert_binary_to_vector(dim).unwrap();
+        assert_eq!(converted.len(), expected.len());
+        for i in 0..3 {
+            assert_eq!(
+                converted.get_ref(i).as_binary().unwrap().unwrap(),
+                expected.get_ref(i).as_binary().unwrap().unwrap()
+            );
+        }
    }
 }
--- a/src/datatypes/src/vectors/eq.rs
+++ b/src/datatypes/src/vectors/eq.rs
@@ -80,7 +80,7 @@ fn equal(lhs: &dyn Vector, rhs: &dyn Vector) -> bool {
    match lhs.data_type() {
        Null(_) => true,
        Boolean(_) => is_vector_eq!(BooleanVector, lhs, rhs),
-        Binary(_) | Json(_) => is_vector_eq!(BinaryVector, lhs, rhs),
+        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),
--- a/src/datatypes/src/vectors/operations.rs
+++ b/src/datatypes/src/vectors/operations.rs
@@ -91,10 +91,17 @@ macro_rules! impl_scalar_vector_op {
            }

            fn cast(&self, to_type: &ConcreteDataType) -> Result<VectorRef> {
-                if to_type == &ConcreteDataType::json_datatype() {
-                    if let Some(vector) = self.as_any().downcast_ref::<BinaryVector>() {
-                        let json_vector = vector.convert_binary_to_json()?;
-                        return Ok(Arc::new(json_vector) as VectorRef);
+                if let Some(vector) = self.as_any().downcast_ref::<BinaryVector>() {
+                    match to_type {
+                        ConcreteDataType::Json(_) => {
+                            let json_vector = vector.convert_binary_to_json()?;
+                            return Ok(Arc::new(json_vector) as VectorRef);
+                        }
+                        ConcreteDataType::Vector(d) => {
+                            let vector = vector.convert_binary_to_vector(d.dim)?;
+                            return Ok(Arc::new(vector) as VectorRef);
+                        }
+                        _ => {}
                    }
                }
                cast::cast_non_constant!(self, to_type)
--- a/src/flow/src/server.rs
+++ b/src/flow/src/server.rs
@@ -37,7 +37,6 @@ use operator::delete::Deleter;
 use operator::insert::Inserter;
 use operator::statement::StatementExecutor;
 use partition::manager::PartitionRuleManager;
-use query::stats::StatementStatistics;
 use query::{QueryEngine, QueryEngineFactory};
 use servers::error::{AlreadyStartedSnafu, StartGrpcSnafu, TcpBindSnafu, TcpIncomingSnafu};
 use servers::server::Server;
@@ -303,7 +302,7 @@ impl FlownodeBuilder {
    ///
    /// or recover all existing flow tasks if in standalone mode(nodeid is None)
    ///
-    /// TODO(discord9): persisent flow tasks with internal state
+    /// TODO(discord9): persistent flow tasks with internal state
    async fn recover_flows(&self, manager: &FlowWorkerManagerRef) -> Result<usize, Error> {
        let nodeid = self.opts.node_id;
        let to_be_recovered: Vec<_> = if let Some(nodeid) = nodeid {
@@ -476,7 +475,6 @@ impl FrontendInvoker {
            layered_cache_registry.clone(),
            inserter.clone(),
            table_route_cache,
-            StatementStatistics::default(),
        ));

        let invoker = FrontendInvoker::new(inserter, deleter, statement_executor);
--- a/src/frontend/Cargo.toml
+++ b/src/frontend/Cargo.toml
@@ -30,6 +30,7 @@ common-function.workspace = true
 common-grpc.workspace = true
 common-macro.workspace = true
 common-meta.workspace = true
+common-options.workspace = true
 common-procedure.workspace = true
 common-query.workspace = true
 common-recordbatch.workspace = true
--- a/src/frontend/src/frontend.rs
+++ b/src/frontend/src/frontend.rs
@@ -13,6 +13,7 @@
 // limitations under the License.

 use common_config::config::Configurable;
+use common_options::datanode::DatanodeClientOptions;
 use common_telemetry::logging::{LoggingOptions, TracingOptions};
 use meta_client::MetaClientOptions;
 use serde::{Deserialize, Serialize};
@@ -22,8 +23,7 @@ use servers::heartbeat_options::HeartbeatOptions;
 use servers::http::HttpOptions;

 use crate::service_config::{
-    DatanodeOptions, InfluxdbOptions, MysqlOptions, OpentsdbOptions, OtlpOptions, PostgresOptions,
-    PromStoreOptions,
+    InfluxdbOptions, MysqlOptions, OpentsdbOptions, OtlpOptions, PostgresOptions, PromStoreOptions,
 };

 #[derive(Clone, Debug, Serialize, Deserialize, PartialEq, Eq)]
@@ -42,7 +42,7 @@ pub struct FrontendOptions {
    pub otlp: OtlpOptions,
    pub meta_client: Option<MetaClientOptions>,
    pub logging: LoggingOptions,
-    pub datanode: DatanodeOptions,
+    pub datanode: DatanodeClientOptions,
    pub user_provider: Option<String>,
    pub export_metrics: ExportMetricsOption,
    pub tracing: TracingOptions,
@@ -64,7 +64,7 @@ impl Default for FrontendOptions {
            otlp: OtlpOptions::default(),
            meta_client: None,
            logging: LoggingOptions::default(),
-            datanode: DatanodeOptions::default(),
+            datanode: DatanodeClientOptions::default(),
            user_provider: None,
            export_metrics: ExportMetricsOption::default(),
            tracing: TracingOptions::default(),
--- a/src/frontend/src/instance.rs
+++ b/src/frontend/src/instance.rs
@@ -51,6 +51,7 @@ use query::metrics::OnDone;
 use query::parser::{PromQuery, QueryLanguageParser, QueryStatement};
 use query::query_engine::options::{validate_catalog_and_schema, QueryOptions};
 use query::query_engine::DescribeResult;
+use query::stats::StatementStatistics;
 use query::QueryEngineRef;
 use raft_engine::{Config, ReadableSize, RecoveryMode};
 use servers::error as server_error;
@@ -122,6 +123,7 @@ pub struct Instance {
    deleter: DeleterRef,
    export_metrics_task: Option<ExportMetricsTask>,
    table_metadata_manager: TableMetadataManagerRef,
+    stats: StatementStatistics,
 }

 impl Instance {
@@ -228,6 +230,10 @@ impl Instance {
        let query_interceptor = self.plugins.get::<SqlQueryInterceptorRef<Error>>();
        let query_interceptor = query_interceptor.as_ref();

+        let _slow_query_timer = self
+            .stats
+            .start_slow_query_timer(QueryStatement::Sql(stmt.clone()));
+
        let output = match stmt {
            Statement::Query(_) | Statement::Explain(_) | Statement::Delete(_) => {
                let stmt = QueryStatement::Sql(stmt);
@@ -412,7 +418,6 @@ impl PrometheusHandler for Instance {
        let interceptor = self
            .plugins
            .get::<PromQueryInterceptorRef<server_error::Error>>();
-        interceptor.pre_execute(query, query_ctx.clone())?;

        self.plugins
            .get::<PermissionCheckerRef>()
@@ -426,9 +431,20 @@ impl PrometheusHandler for Instance {
            }
        })?;

+        let _slow_query_timer = self.stats.start_slow_query_timer(stmt.clone());
+
+        let plan = self
+            .statement_executor
+            .plan(&stmt, query_ctx.clone())
+            .await
+            .map_err(BoxedError::new)
+            .context(ExecuteQuerySnafu)?;
+
+        interceptor.pre_execute(query, Some(&plan), query_ctx.clone())?;
+
        let output = self
            .statement_executor
-            .execute_stmt(stmt, query_ctx.clone())
+            .exec_plan(plan, query_ctx.clone())
            .await
            .map_err(BoxedError::new)
            .context(ExecuteQuerySnafu)?;
--- a/src/frontend/src/instance/builder.rs
+++ b/src/frontend/src/instance/builder.rs
@@ -185,7 +185,6 @@ impl FrontendBuilder {
            local_cache_invalidator,
            inserter.clone(),
            table_route_cache,
-            self.stats,
        ));

        let pipeline_operator = Arc::new(PipelineOperator::new(
@@ -211,6 +210,7 @@ impl FrontendBuilder {
            deleter,
            export_metrics_task: None,
            table_metadata_manager: Arc::new(TableMetadataManager::new(kv_backend)),
+            stats: self.stats,
        })
    }
 }
--- a/src/frontend/src/instance/grpc.rs
+++ b/src/frontend/src/instance/grpc.rs
@@ -19,7 +19,7 @@ use api::v1::{DeleteRequests, DropFlowExpr, InsertRequests, RowDeleteRequests, R
 use async_trait::async_trait;
 use auth::{PermissionChecker, PermissionCheckerRef, PermissionReq};
 use common_query::Output;
-use common_telemetry::tracing;
+use common_telemetry::tracing::{self};
 use query::parser::PromQuery;
 use servers::interceptor::{GrpcQueryInterceptor, GrpcQueryInterceptorRef};
 use servers::query_handler::grpc::GrpcQueryHandler;
--- a/src/frontend/src/instance/otlp.rs
+++ b/src/frontend/src/instance/otlp.rs
@@ -24,7 +24,6 @@ use pipeline::PipelineWay;
 use servers::error::{self, AuthSnafu, Result as ServerResult};
 use servers::interceptor::{OpenTelemetryProtocolInterceptor, OpenTelemetryProtocolInterceptorRef};
 use servers::otlp;
-use servers::otlp::plugin::TraceParserRef;
 use servers::query_handler::OpenTelemetryProtocolHandler;
 use session::context::QueryContextRef;
 use snafu::ResultExt;
@@ -64,6 +63,7 @@ impl OpenTelemetryProtocolHandler for Instance {
    async fn traces(
        &self,
        request: ExportTraceServiceRequest,
+        table_name: String,
        ctx: QueryContextRef,
    ) -> ServerResult<Output> {
        self.plugins
@@ -77,13 +77,7 @@ impl OpenTelemetryProtocolHandler for Instance {
            .get::<OpenTelemetryProtocolInterceptorRef<servers::error::Error>>();
        interceptor_ref.pre_execute(ctx.clone())?;

-        let (table_name, spans) = match self.plugins.get::<TraceParserRef>() {
-            Some(parser) => (parser.table_name(), parser.parse(request)),
-            None => (
-                otlp::trace::TRACE_TABLE_NAME.to_string(),
-                otlp::trace::parse(request),
-            ),
-        };
+        let spans = otlp::trace::parse(request);

        let (requests, rows) = otlp::trace::to_grpc_insert_requests(table_name, spans)?;

--- a/src/frontend/src/service_config.rs
+++ b/src/frontend/src/service_config.rs
@@ -12,7 +12,6 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-pub mod datanode;
 pub mod influxdb;
 pub mod mysql;
 pub mod opentsdb;
@@ -26,5 +25,3 @@ pub use opentsdb::OpentsdbOptions;
 pub use otlp::OtlpOptions;
 pub use postgres::PostgresOptions;
 pub use prom_store::PromStoreOptions;
-
-pub use self::datanode::DatanodeOptions;
--- a/src/index/src/inverted_index/format/reader/blob.rs
+++ b/src/index/src/inverted_index/format/reader/blob.rs
@@ -80,9 +80,7 @@ impl<R: RangeReader> InvertedIndexReader for InvertedIndexBlobReader<R> {
 #[cfg(test)]
 mod tests {
    use common_base::bit_vec::prelude::*;
-    use common_base::range_read::RangeReaderAdapter;
    use fst::MapBuilder;
-    use futures::io::Cursor;
    use greptime_proto::v1::index::{InvertedIndexMeta, InvertedIndexMetas};
    use prost::Message;

@@ -163,8 +161,7 @@ mod tests {
    #[tokio::test]
    async fn test_inverted_index_blob_reader_metadata() {
        let blob = create_inverted_index_blob();
-        let cursor = RangeReaderAdapter(Cursor::new(blob));
-        let mut blob_reader = InvertedIndexBlobReader::new(cursor);
+        let mut blob_reader = InvertedIndexBlobReader::new(blob);

        let metas = blob_reader.metadata().await.unwrap();
        assert_eq!(metas.metas.len(), 2);
@@ -191,8 +188,7 @@ mod tests {
    #[tokio::test]
    async fn test_inverted_index_blob_reader_fst() {
        let blob = create_inverted_index_blob();
-        let cursor = RangeReaderAdapter(Cursor::new(blob));
-        let mut blob_reader = InvertedIndexBlobReader::new(cursor);
+        let mut blob_reader = InvertedIndexBlobReader::new(blob);

        let metas = blob_reader.metadata().await.unwrap();
        let meta = metas.metas.get("tag0").unwrap();
@@ -224,8 +220,7 @@ mod tests {
    #[tokio::test]
    async fn test_inverted_index_blob_reader_bitmap() {
        let blob = create_inverted_index_blob();
-        let cursor = RangeReaderAdapter(Cursor::new(blob));
-        let mut blob_reader = InvertedIndexBlobReader::new(cursor);
+        let mut blob_reader = InvertedIndexBlobReader::new(blob);

        let metas = blob_reader.metadata().await.unwrap();
        let meta = metas.metas.get("tag0").unwrap();
--- a/src/index/src/inverted_index/format/reader/footer.rs
+++ b/src/index/src/inverted_index/format/reader/footer.rs
@@ -113,8 +113,6 @@ impl<R: RangeReader> InvertedIndeFooterReader<R> {

 #[cfg(test)]
 mod tests {
-    use common_base::range_read::RangeReaderAdapter;
-    use futures::io::Cursor;
    use prost::Message;

    use super::*;
@@ -141,10 +139,9 @@ mod tests {
            ..Default::default()
        };

-        let payload_buf = create_test_payload(meta);
+        let mut payload_buf = create_test_payload(meta);
        let blob_size = payload_buf.len() as u64;
-        let cursor = RangeReaderAdapter(Cursor::new(payload_buf));
-        let mut reader = InvertedIndeFooterReader::new(cursor, blob_size);
+        let mut reader = InvertedIndeFooterReader::new(&mut payload_buf, blob_size);

        let payload_size = reader.read_payload_size().await.unwrap();
        let metas = reader.read_payload(payload_size).await.unwrap();
@@ -164,8 +161,7 @@ mod tests {
        let mut payload_buf = create_test_payload(meta);
        payload_buf.push(0xff); // Add an extra byte to corrupt the footer
        let blob_size = payload_buf.len() as u64;
-        let cursor = RangeReaderAdapter(Cursor::new(payload_buf));
-        let mut reader = InvertedIndeFooterReader::new(cursor, blob_size);
+        let mut reader = InvertedIndeFooterReader::new(&mut payload_buf, blob_size);

        let payload_size_result = reader.read_payload_size().await;
        assert!(payload_size_result.is_err());
@@ -180,10 +176,9 @@ mod tests {
            ..Default::default()
        };

-        let payload_buf = create_test_payload(meta);
+        let mut payload_buf = create_test_payload(meta);
        let blob_size = payload_buf.len() as u64;
-        let cursor = RangeReaderAdapter(Cursor::new(payload_buf));
-        let mut reader = InvertedIndeFooterReader::new(cursor, blob_size);
+        let mut reader = InvertedIndeFooterReader::new(&mut payload_buf, blob_size);

        let payload_size = reader.read_payload_size().await.unwrap();
        let payload_result = reader.read_payload(payload_size).await;
--- a/src/index/src/inverted_index/format/writer/blob.rs
+++ b/src/index/src/inverted_index/format/writer/blob.rs
@@ -99,8 +99,6 @@ impl<W: AsyncWrite + Send + Unpin> InvertedIndexBlobWriter<W> {

 #[cfg(test)]
 mod tests {
-    use common_base::range_read::RangeReaderAdapter;
-    use futures::io::Cursor;
    use futures::stream;

    use super::*;
@@ -120,8 +118,7 @@ mod tests {
            .await
            .unwrap();

-        let cursor = RangeReaderAdapter(Cursor::new(blob));
-        let mut reader = InvertedIndexBlobReader::new(cursor);
+        let mut reader = InvertedIndexBlobReader::new(blob);
        let metadata = reader.metadata().await.unwrap();
        assert_eq!(metadata.total_row_count, 8);
        assert_eq!(metadata.segment_row_count, 1);
@@ -161,8 +158,7 @@ mod tests {
            .await
            .unwrap();

-        let cursor = RangeReaderAdapter(Cursor::new(blob));
-        let mut reader = InvertedIndexBlobReader::new(cursor);
+        let mut reader = InvertedIndexBlobReader::new(blob);
        let metadata = reader.metadata().await.unwrap();
        assert_eq!(metadata.total_row_count, 8);
        assert_eq!(metadata.segment_row_count, 1);
--- a/src/meta-client/Cargo.toml
+++ b/src/meta-client/Cargo.toml
@@ -15,6 +15,8 @@ common-grpc.workspace = true
 common-macro.workspace = true
 common-meta.workspace = true
 common-telemetry.workspace = true
+futures.workspace = true
+futures-util.workspace = true
 humantime-serde.workspace = true
 rand.workspace = true
 serde.workspace = true
--- a/src/meta-client/src/client.rs
+++ b/src/meta-client/src/client.rs
@@ -21,6 +21,8 @@ mod cluster;
 mod store;
 mod util;

+use std::sync::Arc;
+
 use api::v1::meta::{ProcedureDetailResponse, Role};
 use cluster::Client as ClusterClient;
 use common_error::ext::BoxedError;
@@ -30,7 +32,8 @@ use common_meta::cluster::{
 };
 use common_meta::datanode::{DatanodeStatKey, DatanodeStatValue, RegionStat};
 use common_meta::ddl::{ExecutorContext, ProcedureExecutor};
-use common_meta::error::{self as meta_error, Result as MetaResult};
+use common_meta::error::{self as meta_error, ExternalSnafu, Result as MetaResult};
+use common_meta::range_stream::PaginationStream;
 use common_meta::rpc::ddl::{SubmitDdlTaskRequest, SubmitDdlTaskResponse};
 use common_meta::rpc::procedure::{
    MigrateRegionRequest, MigrateRegionResponse, ProcedureStateResponse,
@@ -40,8 +43,10 @@ use common_meta::rpc::store::{
    BatchPutResponse, CompareAndPutRequest, CompareAndPutResponse, DeleteRangeRequest,
    DeleteRangeResponse, PutRequest, PutResponse, RangeRequest, RangeResponse,
 };
+use common_meta::rpc::KeyValue;
 use common_meta::ClusterId;
 use common_telemetry::info;
+use futures::TryStreamExt;
 use heartbeat::Client as HeartbeatClient;
 use procedure::Client as ProcedureClient;
 use snafu::{OptionExt, ResultExt};
@@ -314,16 +319,15 @@ impl ClusterInfo for MetaClient {
    }

    async fn list_region_stats(&self) -> Result<Vec<RegionStat>> {
-        let cluster_client = self.cluster_client()?;
+        let cluster_kv_backend = Arc::new(self.cluster_client()?);
        let range_prefix = DatanodeStatKey::key_prefix_with_cluster_id(self.id.0);
        let req = RangeRequest::new().with_prefix(range_prefix);
-        let mut datanode_stats = cluster_client
-            .range(req)
-            .await?
-            .kvs
-            .into_iter()
-            .map(|kv| DatanodeStatValue::try_from(kv.value).context(ConvertMetaRequestSnafu))
-            .collect::<Result<Vec<_>>>()?;
+        let stream = PaginationStream::new(cluster_kv_backend, req, 256, Arc::new(decode_stats))
+            .into_stream();
+        let mut datanode_stats = stream
+            .try_collect::<Vec<_>>()
+            .await
+            .context(ConvertMetaResponseSnafu)?;
        let region_stats = datanode_stats
            .iter_mut()
            .flat_map(|datanode_stat| {
@@ -336,6 +340,12 @@ impl ClusterInfo for MetaClient {
    }
 }

+fn decode_stats(kv: KeyValue) -> MetaResult<DatanodeStatValue> {
+    DatanodeStatValue::try_from(kv.value)
+        .map_err(BoxedError::new)
+        .context(ExternalSnafu)
+}
+
 impl MetaClient {
    pub fn new(id: Id) -> Self {
        Self {
--- a/src/meta-client/src/client/cluster.rs
+++ b/src/meta-client/src/client/cluster.rs
@@ -12,14 +12,22 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

+use std::any::Any;
 use std::future::Future;
 use std::sync::Arc;

 use api::greptime_proto::v1;
 use api::v1::meta::cluster_client::ClusterClient;
 use api::v1::meta::{MetasrvNodeInfo, MetasrvPeersRequest, ResponseHeader, Role};
+use common_error::ext::BoxedError;
 use common_grpc::channel_manager::ChannelManager;
-use common_meta::rpc::store::{BatchGetRequest, BatchGetResponse, RangeRequest, RangeResponse};
+use common_meta::error::{Error as MetaError, ExternalSnafu, Result as MetaResult};
+use common_meta::kv_backend::{KvBackend, TxnService};
+use common_meta::rpc::store::{
+    BatchDeleteRequest, BatchDeleteResponse, BatchGetRequest, BatchGetResponse, BatchPutRequest,
+    BatchPutResponse, DeleteRangeRequest, DeleteRangeResponse, PutRequest, PutResponse,
+    RangeRequest, RangeResponse,
+};
 use common_telemetry::{info, warn};
 use snafu::{ensure, ResultExt};
 use tokio::sync::RwLock;
@@ -79,6 +87,51 @@ impl Client {
    }
 }

+impl TxnService for Client {
+    type Error = MetaError;
+}
+
+#[async_trait::async_trait]
+impl KvBackend for Client {
+    fn name(&self) -> &str {
+        "ClusterClientKvBackend"
+    }
+
+    fn as_any(&self) -> &dyn Any {
+        self
+    }
+
+    async fn range(&self, req: RangeRequest) -> MetaResult<RangeResponse> {
+        self.range(req)
+            .await
+            .map_err(BoxedError::new)
+            .context(ExternalSnafu)
+    }
+
+    async fn put(&self, _: PutRequest) -> MetaResult<PutResponse> {
+        unimplemented!("`put` is not supported in cluster client kv backend")
+    }
+
+    async fn batch_put(&self, _: BatchPutRequest) -> MetaResult<BatchPutResponse> {
+        unimplemented!("`batch_put` is not supported in cluster client kv backend")
+    }
+
+    async fn batch_get(&self, req: BatchGetRequest) -> MetaResult<BatchGetResponse> {
+        self.batch_get(req)
+            .await
+            .map_err(BoxedError::new)
+            .context(ExternalSnafu)
+    }
+
+    async fn delete_range(&self, _: DeleteRangeRequest) -> MetaResult<DeleteRangeResponse> {
+        unimplemented!("`delete_range` is not supported in cluster client kv backend")
+    }
+
+    async fn batch_delete(&self, _: BatchDeleteRequest) -> MetaResult<BatchDeleteResponse> {
+        unimplemented!("`batch_delete` is not supported in cluster client kv backend")
+    }
+}
+
 #[derive(Debug)]
 struct Inner {
    id: Id,
--- a/src/meta-srv/Cargo.toml
+++ b/src/meta-srv/Cargo.toml
@@ -24,6 +24,7 @@ common-greptimedb-telemetry.workspace = true
 common-grpc.workspace = true
 common-macro.workspace = true
 common-meta.workspace = true
+common-options.workspace = true
 common-procedure.workspace = true
 common-runtime.workspace = true
 common-telemetry.workspace = true
--- a/src/meta-srv/src/handler.rs
+++ b/src/meta-srv/src/handler.rs
@@ -478,6 +478,11 @@ pub struct HeartbeatHandlerGroupBuilder {
    /// The handler to handle region lease.
    region_lease_handler: Option<RegionLeaseHandler>,

+    /// The factor that determines how often statistics should be flushed,
+    /// based on the number of received heartbeats. When the number of heartbeats
+    /// reaches this factor, a flush operation is triggered.
+    flush_stats_factor: Option<usize>,
+
    /// The plugins.
    plugins: Option<Plugins>,

@@ -493,6 +498,7 @@ impl HeartbeatHandlerGroupBuilder {
        Self {
            region_failure_handler: None,
            region_lease_handler: None,
+            flush_stats_factor: None,
            plugins: None,
            pushers,
            handlers: vec![],
@@ -510,6 +516,12 @@ impl HeartbeatHandlerGroupBuilder {
        self
    }

+    /// Sets the flush stats factor.
+    pub fn with_flush_stats_factor(mut self, flush_stats_factor: Option<usize>) -> Self {
+        self.flush_stats_factor = flush_stats_factor;
+        self
+    }
+
    /// Sets the [`Plugins`].
    pub fn with_plugins(mut self, plugins: Option<Plugins>) -> Self {
        self.plugins = plugins;
@@ -550,7 +562,7 @@ impl HeartbeatHandlerGroupBuilder {
        if let Some(publish_heartbeat_handler) = publish_heartbeat_handler {
            self.add_handler_last(publish_heartbeat_handler);
        }
-        self.add_handler_last(CollectStatsHandler::default());
+        self.add_handler_last(CollectStatsHandler::new(self.flush_stats_factor));

        self
    }
--- a/src/meta-srv/src/handler/collect_stats_handler.rs
+++ b/src/meta-srv/src/handler/collect_stats_handler.rs
@@ -29,8 +29,6 @@ use crate::error::{self, Result};
 use crate::handler::{HandleControl, HeartbeatAccumulator, HeartbeatHandler};
 use crate::metasrv::Context;

-const MAX_CACHED_STATS_PER_KEY: usize = 10;
-
 #[derive(Debug, Default)]
 struct EpochStats {
    stats: Vec<Stat>,
@@ -69,9 +67,26 @@ impl EpochStats {
    }
 }

-#[derive(Default)]
+const DEFAULT_FLUSH_STATS_FACTOR: usize = 3;
+
 pub struct CollectStatsHandler {
    stats_cache: DashMap<DatanodeStatKey, EpochStats>,
+    flush_stats_factor: usize,
+}
+
+impl Default for CollectStatsHandler {
+    fn default() -> Self {
+        Self::new(None)
+    }
+}
+
+impl CollectStatsHandler {
+    pub fn new(flush_stats_factor: Option<usize>) -> Self {
+        Self {
+            flush_stats_factor: flush_stats_factor.unwrap_or(DEFAULT_FLUSH_STATS_FACTOR),
+            stats_cache: DashMap::default(),
+        }
+    }
 }

 #[async_trait::async_trait]
@@ -130,7 +145,7 @@ impl HeartbeatHandler for CollectStatsHandler {
            rewrite_node_address(ctx, last).await;
        }

-        if !refresh && epoch_stats.len() < MAX_CACHED_STATS_PER_KEY {
+        if !refresh && epoch_stats.len() < self.flush_stats_factor {
            return Ok(HandleControl::Continue);
        }

@@ -261,8 +276,7 @@ mod tests {
        let res = ctx.in_memory.get(&key).await.unwrap();
        let kv = res.unwrap();
        let val: DatanodeStatValue = kv.value.try_into().unwrap();
-        // refresh every 10 stats
-        assert_eq!(10, val.stats.len());
+        assert_eq!(handler.flush_stats_factor, val.stats.len());
    }

    async fn handle_request_many_times(
--- a/src/meta-srv/src/metasrv.rs
+++ b/src/meta-srv/src/metasrv.rs
@@ -24,7 +24,6 @@ use common_base::readable_size::ReadableSize;
 use common_base::Plugins;
 use common_config::Configurable;
 use common_greptimedb_telemetry::GreptimeDBTelemetryTask;
-use common_grpc::channel_manager;
 use common_meta::cache_invalidator::CacheInvalidatorRef;
 use common_meta::ddl::ProcedureExecutorRef;
 use common_meta::key::TableMetadataManagerRef;
@@ -36,6 +35,7 @@ use common_meta::peer::Peer;
 use common_meta::region_keeper::MemoryRegionKeeperRef;
 use common_meta::wal_options_allocator::WalOptionsAllocatorRef;
 use common_meta::{distributed_time_constants, ClusterId};
+use common_options::datanode::DatanodeClientOptions;
 use common_procedure::options::ProcedureConfig;
 use common_procedure::ProcedureManagerRef;
 use common_telemetry::logging::{LoggingOptions, TracingOptions};
@@ -107,7 +107,7 @@ pub struct MetasrvOptions {
    /// The failure detector options.
    pub failure_detector: PhiAccrualFailureDetectorOptions,
    /// The datanode options.
-    pub datanode: DatanodeOptions,
+    pub datanode: DatanodeClientOptions,
    /// Whether to enable telemetry.
    pub enable_telemetry: bool,
    /// The data home directory.
@@ -130,6 +130,10 @@ pub struct MetasrvOptions {
    /// limit the number of operations in a txn because an infinitely large txn could
    /// potentially block other operations.
    pub max_txn_ops: usize,
+    /// The factor that determines how often statistics should be flushed,
+    /// based on the number of received heartbeats. When the number of heartbeats
+    /// reaches this factor, a flush operation is triggered.
+    pub flush_stats_factor: usize,
    /// The tracing options.
    pub tracing: TracingOptions,
    /// The datastore for kv metadata.
@@ -158,13 +162,14 @@ impl Default for MetasrvOptions {
                max_metadata_value_size: Some(ReadableSize::kb(1500)),
            },
            failure_detector: PhiAccrualFailureDetectorOptions::default(),
-            datanode: DatanodeOptions::default(),
+            datanode: DatanodeClientOptions::default(),
            enable_telemetry: true,
            data_home: METASRV_HOME.to_string(),
            wal: MetasrvWalConfig::default(),
            export_metrics: ExportMetricsOption::default(),
            store_key_prefix: String::new(),
            max_txn_ops: 128,
+            flush_stats_factor: 3,
            tracing: TracingOptions::default(),
            backend: BackendImpl::EtcdStore,
        }
@@ -180,35 +185,6 @@ impl Configurable for MetasrvOptions {
 pub struct MetasrvInfo {
    pub server_addr: String,
 }
-
-// Options for datanode.
-#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
-pub struct DatanodeOptions {
-    pub client: DatanodeClientOptions,
-}
-
-// Options for datanode client.
-#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
-pub struct DatanodeClientOptions {
-    #[serde(with = "humantime_serde")]
-    pub timeout: Duration,
-    #[serde(with = "humantime_serde")]
-    pub connect_timeout: Duration,
-    pub tcp_nodelay: bool,
-}
-
-impl Default for DatanodeClientOptions {
-    fn default() -> Self {
-        Self {
-            timeout: Duration::from_secs(channel_manager::DEFAULT_GRPC_REQUEST_TIMEOUT_SECS),
-            connect_timeout: Duration::from_secs(
-                channel_manager::DEFAULT_GRPC_CONNECT_TIMEOUT_SECS,
-            ),
-            tcp_nodelay: true,
-        }
-    }
-}
-
 #[derive(Clone)]
 pub struct Context {
    pub server_addr: String,
--- a/src/meta-srv/src/metasrv/builder.rs
+++ b/src/meta-srv/src/metasrv/builder.rs
@@ -227,7 +227,7 @@ impl MetasrvBuilder {
            ))
        });
        let flow_metadata_allocator = {
-            // for now flownode just use round robin selector
+            // for now flownode just use round-robin selector
            let flow_selector = RoundRobinSelector::new(SelectTarget::Flownode);
            let flow_selector_ctx = selector_ctx.clone();
            let peer_allocator = Arc::new(FlowPeerAllocator::new(
@@ -347,6 +347,7 @@ impl MetasrvBuilder {
                    .with_plugins(plugins.clone())
                    .with_region_failure_handler(region_failover_handler)
                    .with_region_lease_handler(Some(region_lease_handler))
+                    .with_flush_stats_factor(Some(options.flush_stats_factor))
                    .add_default_handlers()
            }
        };
--- a/src/meta-srv/src/selector/common.rs
+++ b/src/meta-srv/src/selector/common.rs
@@ -22,7 +22,7 @@ use crate::metasrv::SelectTarget;
 use crate::selector::SelectorOptions;

 /// According to the `opts`, choose peers from the `weight_array` through `weighted_choose`.
-pub fn choose_peers<W>(opts: &SelectorOptions, weighted_choose: &mut W) -> Result<Vec<Peer>>
+pub fn choose_items<W>(opts: &SelectorOptions, weighted_choose: &mut W) -> Result<Vec<Peer>>
 where
    W: WeightedChoose<Peer>,
 {
@@ -36,20 +36,36 @@ where
        }
    );

-    if opts.allow_duplication {
-        (0..min_required_items)
-            .map(|_| weighted_choose.choose_one())
-            .collect::<Result<_>>()
-    } else {
-        let weight_array_len = weighted_choose.len();
+    if min_required_items == 1 {
+        // fast path
+        return Ok(vec![weighted_choose.choose_one()?]);
+    }

-        // When opts.allow_duplication is false, we need to check that the length of the weighted array is greater than
-        // or equal to min_required_items, otherwise it may cause an infinite loop.
+    let available_count = weighted_choose.len();
+
+    if opts.allow_duplication {
+        // Calculate how many complete rounds of `available_count` items to select,
+        // plus any additional items needed after complete rounds.
+        let complete_batches = min_required_items / available_count;
+        let leftover_items = min_required_items % available_count;
+        if complete_batches == 0 {
+            return weighted_choose.choose_multiple(leftover_items);
+        }
+
+        let mut result = Vec::with_capacity(min_required_items);
+        for _ in 0..complete_batches {
+            result.extend(weighted_choose.choose_multiple(available_count)?);
+        }
+        result.extend(weighted_choose.choose_multiple(leftover_items)?);
+
+        Ok(result)
+    } else {
+        // Ensure the available items are sufficient when duplication is not allowed.
        ensure!(
-            weight_array_len >= min_required_items,
+            available_count >= min_required_items,
            error::NoEnoughAvailableNodeSnafu {
                required: min_required_items,
-                available: weight_array_len,
+                available: available_count,
                select_target: SelectTarget::Datanode
            }
        );
@@ -64,7 +80,7 @@ mod tests {

    use common_meta::peer::Peer;

-    use crate::selector::common::choose_peers;
+    use crate::selector::common::choose_items;
    use crate::selector::weighted_choose::{RandomWeightedChoose, WeightedItem};
    use crate::selector::SelectorOptions;

@@ -115,7 +131,7 @@ mod tests {
            };

            let selected_peers: HashSet<_> =
-                choose_peers(&opts, &mut RandomWeightedChoose::new(weight_array.clone()))
+                choose_items(&opts, &mut RandomWeightedChoose::new(weight_array.clone()))
                    .unwrap()
                    .into_iter()
                    .collect();
@@ -129,7 +145,7 @@ mod tests {
        };

        let selected_result =
-            choose_peers(&opts, &mut RandomWeightedChoose::new(weight_array.clone()));
+            choose_items(&opts, &mut RandomWeightedChoose::new(weight_array.clone()));
        assert!(selected_result.is_err());

        for i in 1..=50 {
@@ -139,7 +155,7 @@ mod tests {
            };

            let selected_peers =
-                choose_peers(&opts, &mut RandomWeightedChoose::new(weight_array.clone())).unwrap();
+                choose_items(&opts, &mut RandomWeightedChoose::new(weight_array.clone())).unwrap();

            assert_eq!(i, selected_peers.len());
        }
--- a/src/meta-srv/src/selector/lease_based.rs
+++ b/src/meta-srv/src/selector/lease_based.rs
@@ -17,7 +17,7 @@ use common_meta::peer::Peer;
 use crate::error::Result;
 use crate::lease;
 use crate::metasrv::SelectorContext;
-use crate::selector::common::choose_peers;
+use crate::selector::common::choose_items;
 use crate::selector::weighted_choose::{RandomWeightedChoose, WeightedItem};
 use crate::selector::{Namespace, Selector, SelectorOptions};

@@ -53,7 +53,7 @@ impl Selector for LeaseBasedSelector {

        // 3. choose peers by weight_array.
        let mut weighted_choose = RandomWeightedChoose::new(weight_array);
-        let selected = choose_peers(&opts, &mut weighted_choose)?;
+        let selected = choose_items(&opts, &mut weighted_choose)?;

        Ok(selected)
    }
--- a/src/meta-srv/src/selector/load_based.rs
+++ b/src/meta-srv/src/selector/load_based.rs
@@ -26,7 +26,7 @@ use crate::error::{self, Result};
 use crate::key::{DatanodeLeaseKey, LeaseValue};
 use crate::lease;
 use crate::metasrv::SelectorContext;
-use crate::selector::common::choose_peers;
+use crate::selector::common::choose_items;
 use crate::selector::weight_compute::{RegionNumsBasedWeightCompute, WeightCompute};
 use crate::selector::weighted_choose::RandomWeightedChoose;
 use crate::selector::{Namespace, Selector, SelectorOptions};
@@ -94,7 +94,7 @@ where

        // 5. choose peers by weight_array.
        let mut weighted_choose = RandomWeightedChoose::new(weight_array);
-        let selected = choose_peers(&opts, &mut weighted_choose)?;
+        let selected = choose_items(&opts, &mut weighted_choose)?;

        debug!(
            "LoadBasedSelector select peers: {:?}, namespace: {}, opts: {:?}.",
--- a/src/metric-engine/src/data_region.rs
+++ b/src/metric-engine/src/data_region.rs
@@ -15,8 +15,7 @@
 use api::v1::SemanticType;
 use common_error::ext::ErrorExt;
 use common_error::status_code::StatusCode;
-use common_telemetry::info;
-use common_telemetry::tracing::warn;
+use common_telemetry::{debug, info, warn};
 use mito2::engine::MitoEngine;
 use snafu::ResultExt;
 use store_api::metadata::ColumnMetadata;
@@ -150,6 +149,7 @@ impl DataRegion {
            })
            .collect::<Result<_>>()?;

+        debug!("Adding (Column id assigned) columns {new_columns:?} to region {region_id:?}");
        // assemble alter request
        let alter_request = RegionRequest::Alter(RegionAlterRequest {
            schema_version: version,
--- a/src/metric-engine/src/engine/alter.rs
+++ b/src/metric-engine/src/engine/alter.rs
@@ -92,17 +92,29 @@ impl MetricEngineInner {

        let metadata_region_id = to_metadata_region_id(physical_region_id);
        let mut columns_to_add = vec![];
+        // columns that already exist in physical region
+        let mut existing_columns = vec![];
+
+        let pre_existing_physical_columns = self
+            .data_region
+            .physical_columns(physical_region_id)
+            .await?;
+
+        let pre_exist_cols = pre_existing_physical_columns
+            .iter()
+            .map(|col| (col.column_schema.name.as_str(), col))
+            .collect::<HashMap<_, _>>();
+
+        // check pre-existing physical columns so if any columns to add is already exist,
+        // we can skip it in physical alter operation
+        // (but still need to update them in logical alter operation)
        for col in &columns {
-            if self
-                .metadata_region
-                .column_semantic_type(
-                    metadata_region_id,
-                    logical_region_id,
-                    &col.column_metadata.column_schema.name,
-                )
-                .await?
-                .is_none()
+            if let Some(exist_column) =
+                pre_exist_cols.get(&col.column_metadata.column_schema.name.as_str())
            {
+                // push the correct column schema with correct column id
+                existing_columns.push(*exist_column);
+            } else {
                columns_to_add.push(col.column_metadata.clone());
            }
        }
@@ -111,16 +123,16 @@ impl MetricEngineInner {
        let data_region_id = to_data_region_id(physical_region_id);
        self.add_columns_to_physical_data_region(
            data_region_id,
-            metadata_region_id,
            logical_region_id,
-            columns_to_add,
+            &mut columns_to_add,
        )
        .await?;

-        // register columns to logical region
-        for col in columns {
+        // note here we don't use `columns` directly but concat `existing_columns` with `columns_to_add` to get correct metadata
+        // about already existing columns
+        for metadata in existing_columns.into_iter().chain(columns_to_add.iter()) {
            self.metadata_region
-                .add_column(metadata_region_id, logical_region_id, &col.column_metadata)
+                .add_column(metadata_region_id, logical_region_id, metadata)
                .await?;
        }

--- a/src/metric-engine/src/engine/create.rs
+++ b/src/metric-engine/src/engine/create.rs
@@ -16,7 +16,7 @@ use std::collections::{HashMap, HashSet};

 use api::v1::SemanticType;
 use common_error::ext::BoxedError;
-use common_telemetry::info;
+use common_telemetry::{info, warn};
 use common_time::Timestamp;
 use datatypes::data_type::ConcreteDataType;
 use datatypes::schema::ColumnSchema;
@@ -31,7 +31,6 @@ use store_api::metric_engine_consts::{
    METADATA_SCHEMA_KEY_COLUMN_INDEX, METADATA_SCHEMA_KEY_COLUMN_NAME,
    METADATA_SCHEMA_TIMESTAMP_COLUMN_INDEX, METADATA_SCHEMA_TIMESTAMP_COLUMN_NAME,
    METADATA_SCHEMA_VALUE_COLUMN_INDEX, METADATA_SCHEMA_VALUE_COLUMN_NAME,
-    PHYSICAL_TABLE_METADATA_KEY,
 };
 use store_api::mito_engine_options::{APPEND_MODE_KEY, TTL_KEY};
 use store_api::region_engine::RegionEngine;
@@ -61,7 +60,7 @@ impl MetricEngineInner {
    ) -> Result<AffectedRows> {
        Self::verify_region_create_request(&request)?;

-        let result = if request.options.contains_key(PHYSICAL_TABLE_METADATA_KEY) {
+        let result = if request.is_physical_table() {
            self.create_physical_region(region_id, request).await
        } else if request.options.contains_key(LOGICAL_TABLE_METADATA_KEY) {
            let physical_region_id = self.create_logical_region(region_id, request).await?;
@@ -212,11 +211,17 @@ impl MetricEngineInner {

            self.add_columns_to_physical_data_region(
                data_region_id,
-                metadata_region_id,
                logical_region_id,
-                new_columns,
+                &mut new_columns,
            )
            .await?;
+
+            // register columns to metadata region
+            for col in &new_columns {
+                self.metadata_region
+                    .add_column(metadata_region_id, logical_region_id, col)
+                    .await?;
+            }
        }

        // register logical region to metadata region
@@ -260,27 +265,24 @@ impl MetricEngineInner {
        Ok(data_region_id)
    }

-    /// Execute corresponding alter requests to mito region. New added columns' [ColumnMetadata] will be
-    /// cloned into `added_columns`.
+    /// Execute corresponding alter requests to mito region. After calling this, `new_columns` will be assign a new column id
+    /// which should be correct if the following requirements are met:
+    ///
+    /// # NOTE
+    ///
+    /// `new_columns` MUST NOT pre-exist in the physical region. Or the results will be wrong column id for the new columns.
+    ///
    pub(crate) async fn add_columns_to_physical_data_region(
        &self,
        data_region_id: RegionId,
-        metadata_region_id: RegionId,
        logical_region_id: RegionId,
-        mut new_columns: Vec<ColumnMetadata>,
+        new_columns: &mut [ColumnMetadata],
    ) -> Result<()> {
        // alter data region
        self.data_region
-            .add_columns(data_region_id, &mut new_columns)
+            .add_columns(data_region_id, new_columns)
            .await?;

-        // register columns to metadata region
-        for col in &new_columns {
-            self.metadata_region
-                .add_column(metadata_region_id, logical_region_id, col)
-                .await?;
-        }
-
        // safety: previous step has checked this
        self.state.write().unwrap().add_physical_columns(
            data_region_id,
@@ -291,6 +293,34 @@ impl MetricEngineInner {
        info!("Create region {logical_region_id} leads to adding columns {new_columns:?} to physical region {data_region_id}");
        PHYSICAL_COLUMN_COUNT.add(new_columns.len() as _);

+        // correct the column id
+        let after_alter_physical_schema = self.data_region.physical_columns(data_region_id).await?;
+        let after_alter_physical_schema_map = after_alter_physical_schema
+            .iter()
+            .map(|metadata| (metadata.column_schema.name.as_str(), metadata))
+            .collect::<HashMap<_, _>>();
+
+        // double check to make sure column ids are not mismatched
+        // shouldn't be a expensive operation, given it only query for physical columns
+        for col in new_columns.iter_mut() {
+            let column_metadata = after_alter_physical_schema_map
+                .get(&col.column_schema.name.as_str())
+                .with_context(|| ColumnNotFoundSnafu {
+                    name: &col.column_schema.name,
+                    region_id: data_region_id,
+                })?;
+            if col != *column_metadata {
+                warn!(
+                    "Add already existing columns with different column metadata to physical region({:?}): new column={:?}, old column={:?}", 
+                    data_region_id,
+                    col,
+                    column_metadata
+                );
+                // update to correct metadata
+                *col = (*column_metadata).clone();
+            }
+        }
+
        Ok(())
    }

@@ -324,12 +354,11 @@ impl MetricEngineInner {

        // check if required table option is present
        ensure!(
-            request.options.contains_key(PHYSICAL_TABLE_METADATA_KEY)
-                || request.options.contains_key(LOGICAL_TABLE_METADATA_KEY),
+            request.is_physical_table() || request.options.contains_key(LOGICAL_TABLE_METADATA_KEY),
            MissingRegionOptionSnafu {}
        );
        ensure!(
-            !(request.options.contains_key(PHYSICAL_TABLE_METADATA_KEY)
+            !(request.is_physical_table()
                && request.options.contains_key(LOGICAL_TABLE_METADATA_KEY)),
            ConflictRegionOptionSnafu {}
        );
@@ -512,7 +541,7 @@ impl MetricEngineInner {

 #[cfg(test)]
 mod test {
-    use store_api::metric_engine_consts::METRIC_ENGINE_NAME;
+    use store_api::metric_engine_consts::{METRIC_ENGINE_NAME, PHYSICAL_TABLE_METADATA_KEY};

    use super::*;
    use crate::engine::MetricEngine;
--- a/src/metric-engine/src/engine/open.rs
+++ b/src/metric-engine/src/engine/open.rs
@@ -18,9 +18,7 @@ use common_telemetry::info;
 use mito2::engine::MITO_ENGINE_NAME;
 use object_store::util::join_dir;
 use snafu::ResultExt;
-use store_api::metric_engine_consts::{
-    DATA_REGION_SUBDIR, METADATA_REGION_SUBDIR, PHYSICAL_TABLE_METADATA_KEY,
-};
+use store_api::metric_engine_consts::{DATA_REGION_SUBDIR, METADATA_REGION_SUBDIR};
 use store_api::region_engine::RegionEngine;
 use store_api::region_request::{AffectedRows, RegionOpenRequest, RegionRequest};
 use store_api::storage::RegionId;
@@ -46,9 +44,7 @@ impl MetricEngineInner {
        region_id: RegionId,
        request: RegionOpenRequest,
    ) -> Result<AffectedRows> {
-        let is_opening_physical_region = request.options.contains_key(PHYSICAL_TABLE_METADATA_KEY);
-
-        if is_opening_physical_region {
+        if request.is_physical_table() {
            // open physical region and recover states
            self.open_physical_region(region_id, request).await?;
            self.recover_states(region_id).await?;
--- a/src/metric-engine/src/engine/region_metadata.rs
+++ b/src/metric-engine/src/engine/region_metadata.rs
@@ -14,6 +14,8 @@

 //! Implementation of retrieving logical region's region metadata.

+use std::collections::HashMap;
+
 use store_api::metadata::ColumnMetadata;
 use store_api::storage::RegionId;

@@ -46,23 +48,36 @@ impl MetricEngineInner {
            .read_lock_logical_region(logical_region_id)
            .await;
        // Load logical and physical columns, and intersect them to get logical column metadata.
-        let mut logical_column_metadata = self
+        let logical_column_metadata = self
            .metadata_region
            .logical_columns(physical_region_id, logical_region_id)
            .await?
            .into_iter()
            .map(|(_, column_metadata)| column_metadata)
            .collect::<Vec<_>>();
-        // Sort columns on column name to ensure the order
-        logical_column_metadata
-            .sort_unstable_by(|c1, c2| c1.column_schema.name.cmp(&c2.column_schema.name));
-        // Update cache
-        self.state
-            .write()
-            .unwrap()
-            .add_logical_columns(logical_region_id, logical_column_metadata.clone());

-        Ok(logical_column_metadata)
+        // Update cache
+        let mut mutable_state = self.state.write().unwrap();
+        // Merge with existing cached columns.
+        let existing_columns = mutable_state
+            .logical_columns()
+            .get(&logical_region_id)
+            .cloned()
+            .unwrap_or_default()
+            .into_iter();
+        let mut dedup_columns = logical_column_metadata
+            .into_iter()
+            .chain(existing_columns)
+            .map(|c| (c.column_id, c))
+            .collect::<HashMap<_, _>>()
+            .values()
+            .cloned()
+            .collect::<Vec<_>>();
+        // Sort columns on column name to ensure the order
+        dedup_columns.sort_unstable_by(|c1, c2| c1.column_schema.name.cmp(&c2.column_schema.name));
+        mutable_state.set_logical_columns(logical_region_id, dedup_columns.clone());
+
+        Ok(dedup_columns)
    }

    /// Load logical column names of a logical region.
--- a/src/metric-engine/src/engine/state.rs
+++ b/src/metric-engine/src/engine/state.rs
@@ -85,19 +85,13 @@ impl MetricEngineState {
            .insert(logical_region_id, physical_region_id);
    }

-    /// Add and reorder logical columns.
-    ///
-    /// Caller should make sure:
-    /// 1. there is no duplicate columns
-    /// 2. the column order is the same with the order in the metadata, which is
-    ///    alphabetically ordered on column name.
-    pub fn add_logical_columns(
+    /// Replace the logical columns of the logical region with given columns.
+    pub fn set_logical_columns(
        &mut self,
        logical_region_id: RegionId,
-        new_columns: impl IntoIterator<Item = ColumnMetadata>,
+        columns: Vec<ColumnMetadata>,
    ) {
-        let columns = self.logical_columns.entry(logical_region_id).or_default();
-        columns.extend(new_columns);
+        self.logical_columns.insert(logical_region_id, columns);
    }

    pub fn get_physical_region_id(&self, logical_region_id: RegionId) -> Option<RegionId> {
--- a/src/metric-engine/src/metadata_region.rs
+++ b/src/metric-engine/src/metadata_region.rs
@@ -211,6 +211,7 @@ impl MetadataRegion {
    }

    /// Check if the given column exists. Return the semantic type if exists.
+    #[cfg(test)]
    pub async fn column_semantic_type(
        &self,
        physical_region_id: RegionId,
@@ -373,6 +374,7 @@ impl MetadataRegion {

    /// Retrieves the value associated with the given key in the specified region.
    /// Returns `Ok(None)` if the key is not found.
+    #[cfg(test)]
    pub async fn get(&self, region_id: RegionId, key: &str) -> Result<Option<String>> {
        let scan_req = Self::build_read_request(key);
        let record_batch_stream = self
--- a/src/mito2/Cargo.toml
+++ b/src/mito2/Cargo.toml
@@ -24,6 +24,7 @@ common-datasource.workspace = true
 common-decimal.workspace = true
 common-error.workspace = true
 common-macro.workspace = true
+common-meta.workspace = true
 common-query.workspace = true
 common-recordbatch.workspace = true
 common-runtime.workspace = true
@@ -74,6 +75,7 @@ uuid.workspace = true

 [dev-dependencies]
 common-function.workspace = true
+common-meta = { workspace = true, features = ["testing"] }
 common-procedure-test.workspace = true
 common-test-util.workspace = true
 criterion = "0.4"
--- a/src/mito2/src/compaction.rs
+++ b/src/mito2/src/compaction.rs
@@ -28,7 +28,8 @@ use std::time::{Duration, Instant};

 use api::v1::region::compact_request;
 use common_base::Plugins;
-use common_telemetry::{debug, error, info};
+use common_meta::key::SchemaMetadataManagerRef;
+use common_telemetry::{debug, error, info, warn};
 use common_time::range::TimestampRange;
 use common_time::timestamp::TimeUnit;
 use common_time::Timestamp;
@@ -37,7 +38,7 @@ use datafusion_expr::Expr;
 use serde::{Deserialize, Serialize};
 use snafu::{OptionExt, ResultExt};
 use store_api::metadata::RegionMetadataRef;
-use store_api::storage::RegionId;
+use store_api::storage::{RegionId, TableId};
 use table::predicate::Predicate;
 use tokio::sync::mpsc::{self, Sender};

@@ -48,8 +49,8 @@ use crate::compaction::picker::{new_picker, CompactionTask};
 use crate::compaction::task::CompactionTaskImpl;
 use crate::config::MitoConfig;
 use crate::error::{
-    CompactRegionSnafu, Error, RegionClosedSnafu, RegionDroppedSnafu, RegionTruncatedSnafu,
-    RemoteCompactionSnafu, Result, TimeRangePredicateOverflowSnafu,
+    CompactRegionSnafu, Error, GetSchemaMetadataSnafu, RegionClosedSnafu, RegionDroppedSnafu,
+    RegionTruncatedSnafu, RemoteCompactionSnafu, Result, TimeRangePredicateOverflowSnafu,
 };
 use crate::metrics::COMPACTION_STAGE_ELAPSED;
 use crate::read::projection::ProjectionMapper;
@@ -82,6 +83,7 @@ pub struct CompactionRequest {
    pub(crate) cache_manager: CacheManagerRef,
    pub(crate) manifest_ctx: ManifestContextRef,
    pub(crate) listener: WorkerListener,
+    pub(crate) schema_metadata_manager: SchemaMetadataManagerRef,
 }

 impl CompactionRequest {
@@ -141,6 +143,7 @@ impl CompactionScheduler {
        access_layer: &AccessLayerRef,
        waiter: OptionOutputTx,
        manifest_ctx: &ManifestContextRef,
+        schema_metadata_manager: SchemaMetadataManagerRef,
    ) -> Result<()> {
        if let Some(status) = self.region_status.get_mut(&region_id) {
            // Region is compacting. Add the waiter to pending list.
@@ -158,6 +161,7 @@ impl CompactionScheduler {
            self.cache_manager.clone(),
            manifest_ctx,
            self.listener.clone(),
+            schema_metadata_manager,
        );
        self.region_status.insert(region_id, status);
        let result = self
@@ -173,6 +177,7 @@ impl CompactionScheduler {
        &mut self,
        region_id: RegionId,
        manifest_ctx: &ManifestContextRef,
+        schema_metadata_manager: SchemaMetadataManagerRef,
    ) {
        let Some(status) = self.region_status.get_mut(&region_id) else {
            return;
@@ -186,6 +191,7 @@ impl CompactionScheduler {
            self.cache_manager.clone(),
            manifest_ctx,
            self.listener.clone(),
+            schema_metadata_manager,
        );
        // Try to schedule next compaction task for this region.
        if let Err(e) = self
@@ -256,10 +262,23 @@ impl CompactionScheduler {
            cache_manager,
            manifest_ctx,
            listener,
+            schema_metadata_manager,
        } = request;
+
+        let ttl = find_ttl(
+            region_id.table_id(),
+            current_version.options.ttl,
+            &schema_metadata_manager,
+        )
+        .await
+        .unwrap_or_else(|e| {
+            warn!(e; "Failed to get ttl for region: {}", region_id);
+            None
+        });
+
        debug!(
-            "Pick compaction strategy {:?} for region: {}",
-            picker, region_id
+            "Pick compaction strategy {:?} for region: {}, ttl: {:?}",
+            picker, region_id, ttl
        );

        let compaction_region = CompactionRegion {
@@ -273,6 +292,7 @@ impl CompactionScheduler {
            access_layer: access_layer.clone(),
            manifest_ctx: manifest_ctx.clone(),
            file_purger: None,
+            ttl,
        };

        let picker_output = {
@@ -414,6 +434,24 @@ impl PendingCompaction {
    }
 }

+/// Finds TTL of table by first examine table options then database options.
+async fn find_ttl(
+    table_id: TableId,
+    table_ttl: Option<Duration>,
+    schema_metadata_manager: &SchemaMetadataManagerRef,
+) -> Result<Option<Duration>> {
+    if let Some(table_ttl) = table_ttl {
+        return Ok(Some(table_ttl));
+    }
+
+    let ttl = schema_metadata_manager
+        .get_schema_options_by_table_id(table_id)
+        .await
+        .context(GetSchemaMetadataSnafu)?
+        .and_then(|options| options.ttl);
+    Ok(ttl)
+}
+
 /// Status of running and pending region compaction tasks.
 struct CompactionStatus {
    /// Id of the region.
@@ -471,6 +509,7 @@ impl CompactionStatus {
        cache_manager: CacheManagerRef,
        manifest_ctx: &ManifestContextRef,
        listener: WorkerListener,
+        schema_metadata_manager: SchemaMetadataManagerRef,
    ) -> CompactionRequest {
        let current_version = self.version_control.current().version;
        let start_time = Instant::now();
@@ -484,6 +523,7 @@ impl CompactionStatus {
            cache_manager,
            manifest_ctx: manifest_ctx.clone(),
            listener,
+            schema_metadata_manager,
        };

        if let Some(pending) = self.pending_compaction.take() {
@@ -639,6 +679,9 @@ fn get_expired_ssts(

 #[cfg(test)]
 mod tests {
+    use common_meta::key::SchemaMetadataManager;
+    use common_meta::kv_backend::memory::MemoryKvBackend;
+    use common_meta::kv_backend::KvBackendRef;
    use tokio::sync::oneshot;

    use super::*;
@@ -651,7 +694,19 @@ mod tests {
        let (tx, _rx) = mpsc::channel(4);
        let mut scheduler = env.mock_compaction_scheduler(tx);
        let mut builder = VersionControlBuilder::new();
-
+        let schema_metadata_manager = Arc::new(SchemaMetadataManager::new(Arc::new(
+            MemoryKvBackend::new(),
+        )
+            as KvBackendRef));
+        schema_metadata_manager
+            .register_region_table_info(
+                builder.region_id().table_id(),
+                "test_table",
+                "test_catalog",
+                "test_schema",
+                None,
+            )
+            .await;
        // Nothing to compact.
        let version_control = Arc::new(builder.build());
        let (output_tx, output_rx) = oneshot::channel();
@@ -667,6 +722,7 @@ mod tests {
                &env.access_layer,
                waiter,
                &manifest_ctx,
+                schema_metadata_manager.clone(),
            )
            .await
            .unwrap();
@@ -686,6 +742,7 @@ mod tests {
                &env.access_layer,
                waiter,
                &manifest_ctx,
+                schema_metadata_manager,
            )
            .await
            .unwrap();
@@ -703,6 +760,19 @@ mod tests {
        let mut builder = VersionControlBuilder::new();
        let purger = builder.file_purger();
        let region_id = builder.region_id();
+        let schema_metadata_manager = Arc::new(SchemaMetadataManager::new(Arc::new(
+            MemoryKvBackend::new(),
+        )
+            as KvBackendRef));
+        schema_metadata_manager
+            .register_region_table_info(
+                builder.region_id().table_id(),
+                "test_table",
+                "test_catalog",
+                "test_schema",
+                None,
+            )
+            .await;

        // 5 files to compact.
        let end = 1000 * 1000;
@@ -726,6 +796,7 @@ mod tests {
                &env.access_layer,
                OptionOutputTx::none(),
                &manifest_ctx,
+                schema_metadata_manager.clone(),
            )
            .await
            .unwrap();
@@ -755,6 +826,7 @@ mod tests {
                &env.access_layer,
                OptionOutputTx::none(),
                &manifest_ctx,
+                schema_metadata_manager.clone(),
            )
            .await
            .unwrap();
@@ -769,7 +841,7 @@ mod tests {

        // On compaction finished and schedule next compaction.
        scheduler
-            .on_compaction_finished(region_id, &manifest_ctx)
+            .on_compaction_finished(region_id, &manifest_ctx, schema_metadata_manager.clone())
            .await;
        assert_eq!(1, scheduler.region_status.len());
        assert_eq!(2, job_scheduler.num_jobs());
@@ -789,6 +861,7 @@ mod tests {
                &env.access_layer,
                OptionOutputTx::none(),
                &manifest_ctx,
+                schema_metadata_manager,
            )
            .await
            .unwrap();
--- a/src/mito2/src/compaction/compactor.rs
+++ b/src/mito2/src/compaction/compactor.rs
@@ -16,7 +16,8 @@ use std::sync::Arc;
 use std::time::Duration;

 use api::v1::region::compact_request;
-use common_telemetry::info;
+use common_meta::key::SchemaMetadataManagerRef;
+use common_telemetry::{info, warn};
 use object_store::manager::ObjectStoreManagerRef;
 use serde::{Deserialize, Serialize};
 use smallvec::SmallVec;
@@ -27,7 +28,7 @@ use store_api::storage::RegionId;
 use crate::access_layer::{AccessLayer, AccessLayerRef, OperationType, SstWriteRequest};
 use crate::cache::{CacheManager, CacheManagerRef};
 use crate::compaction::picker::{new_picker, PickerOutput};
-use crate::compaction::CompactionSstReaderBuilder;
+use crate::compaction::{find_ttl, CompactionSstReaderBuilder};
 use crate::config::MitoConfig;
 use crate::error::{EmptyRegionDirSnafu, JoinSnafu, ObjectStoreNotFoundSnafu, Result};
 use crate::manifest::action::{RegionEdit, RegionMetaAction, RegionMetaActionList};
@@ -62,6 +63,7 @@ pub struct CompactionRegion {
    pub(crate) manifest_ctx: Arc<ManifestContext>,
    pub(crate) current_version: VersionRef,
    pub(crate) file_purger: Option<Arc<LocalFilePurger>>,
+    pub(crate) ttl: Option<Duration>,
 }

 /// OpenCompactionRegionRequest represents the request to open a compaction region.
@@ -78,6 +80,7 @@ pub async fn open_compaction_region(
    req: &OpenCompactionRegionRequest,
    mito_config: &MitoConfig,
    object_store_manager: ObjectStoreManagerRef,
+    schema_metadata_manager: SchemaMetadataManagerRef,
 ) -> Result<CompactionRegion> {
    let object_store = {
        let name = &req.region_options.storage;
@@ -169,6 +172,16 @@ pub async fn open_compaction_region(
        Arc::new(version)
    };

+    let ttl = find_ttl(
+        req.region_id.table_id(),
+        current_version.options.ttl,
+        &schema_metadata_manager,
+    )
+    .await
+    .unwrap_or_else(|e| {
+        warn!(e; "Failed to get ttl for region: {}", region_metadata.region_id);
+        None
+    });
    Ok(CompactionRegion {
        region_id: req.region_id,
        region_options: req.region_options.clone(),
@@ -180,6 +193,7 @@ pub async fn open_compaction_region(
        manifest_ctx,
        current_version,
        file_purger: Some(file_purger),
+        ttl,
    })
 }

--- a/src/mito2/src/compaction/twcs.rs
+++ b/src/mito2/src/compaction/twcs.rs
@@ -212,8 +212,9 @@ impl Picker for TwcsPicker {
    fn pick(&self, compaction_region: &CompactionRegion) -> Option<PickerOutput> {
        let region_id = compaction_region.region_id;
        let levels = compaction_region.current_version.ssts.levels();
-        let ttl = compaction_region.current_version.options.ttl;
-        let expired_ssts = get_expired_ssts(levels, ttl, Timestamp::current_millis());
+
+        let expired_ssts =
+            get_expired_ssts(levels, compaction_region.ttl, Timestamp::current_millis());
        if !expired_ssts.is_empty() {
            info!("Expired SSTs in region {}: {:?}", region_id, expired_ssts);
            // here we mark expired SSTs as compacting to avoid them being picked.
@@ -297,6 +298,9 @@ fn assign_to_windows<'a>(
    let mut windows: HashMap<i64, Window> = HashMap::new();
    // Iterates all files and assign to time windows according to max timestamp
    for f in files {
+        if f.compacting() {
+            continue;
+        }
        let (_, end) = f.time_range();
        let time_window = end
            .convert_to(TimeUnit::Second)
@@ -443,6 +447,21 @@ mod tests {
        );
    }

+    #[test]
+    fn test_assign_compacting_to_windows() {
+        let files = [
+            new_file_handle(FileId::random(), 0, 999, 0),
+            new_file_handle(FileId::random(), 0, 999, 0),
+            new_file_handle(FileId::random(), 0, 999, 0),
+            new_file_handle(FileId::random(), 0, 999, 0),
+            new_file_handle(FileId::random(), 0, 999, 0),
+        ];
+        files[0].set_compacting(true);
+        files[2].set_compacting(true);
+        let windows = assign_to_windows(files.iter(), 3);
+        assert_eq!(3, windows.get(&0).unwrap().files.len());
+    }
+
    /// (Window value, overlapping, files' time ranges in window)
    type ExpectedWindowSpec = (i64, bool, Vec<(i64, i64)>);

--- a/src/mito2/src/compaction/window.rs
+++ b/src/mito2/src/compaction/window.rs
@@ -352,6 +352,28 @@ mod tests {
        );
    }

+    #[test]
+    fn test_assign_compacting_files_to_windows() {
+        let picker = WindowedCompactionPicker::new(Some(HOUR / 1000));
+        let files = vec![
+            (FileId::random(), 0, 2 * HOUR - 1, 0),
+            (FileId::random(), HOUR, HOUR * 3 - 1, 0),
+        ];
+        let version = build_version(&files, Some(Duration::from_millis(3 * HOUR as u64)));
+        version.ssts.levels()[0]
+            .files()
+            .for_each(|f| f.set_compacting(true));
+        let (outputs, expired_ssts, window_seconds) = picker.pick_inner(
+            RegionId::new(0, 0),
+            &version,
+            Timestamp::new_millisecond(HOUR * 3),
+        );
+
+        assert!(expired_ssts.is_empty());
+        assert_eq!(HOUR / 1000, window_seconds);
+        assert!(outputs.is_empty());
+    }
+
    #[test]
    fn test_file_time_bucket_span() {
        assert_eq!(
--- a/src/mito2/src/config.rs
+++ b/src/mito2/src/config.rs
@@ -150,7 +150,7 @@ impl Default for MitoConfig {
            selector_result_cache_size: ReadableSize::mb(512),
            enable_experimental_write_cache: false,
            experimental_write_cache_path: String::new(),
-            experimental_write_cache_size: ReadableSize::mb(512),
+            experimental_write_cache_size: ReadableSize::gb(1),
            experimental_write_cache_ttl: None,
            sst_write_buffer_size: DEFAULT_WRITE_BUFFER_SIZE,
            scan_parallelism: divide_num_cpus(4),
--- a/src/mito2/src/engine.rs
+++ b/src/mito2/src/engine.rs
@@ -66,6 +66,7 @@ use api::region::RegionResponse;
 use async_trait::async_trait;
 use common_base::Plugins;
 use common_error::ext::BoxedError;
+use common_meta::key::SchemaMetadataManagerRef;
 use common_recordbatch::SendableRecordBatchStream;
 use common_telemetry::tracing;
 use common_wal::options::{WalOptions, WAL_OPTIONS_KEY};
@@ -89,7 +90,7 @@ use crate::error::{
 };
 use crate::manifest::action::RegionEdit;
 use crate::metrics::HANDLE_REQUEST_ELAPSED;
-use crate::read::scan_region::{ScanParallism, ScanRegion, Scanner};
+use crate::read::scan_region::{ScanParallelism, ScanRegion, Scanner};
 use crate::request::{RegionEditRequest, WorkerRequest};
 use crate::wal::entry_distributor::{
    build_wal_entry_distributor_and_receivers, DEFAULT_ENTRY_RECEIVER_BUFFER_SIZE,
@@ -112,13 +113,21 @@ impl MitoEngine {
        mut config: MitoConfig,
        log_store: Arc<S>,
        object_store_manager: ObjectStoreManagerRef,
+        schema_metadata_manager: SchemaMetadataManagerRef,
        plugins: Plugins,
    ) -> Result<MitoEngine> {
        config.sanitize(data_home)?;

        Ok(MitoEngine {
            inner: Arc::new(
-                EngineInner::new(config, log_store, object_store_manager, plugins).await?,
+                EngineInner::new(
+                    config,
+                    log_store,
+                    object_store_manager,
+                    schema_metadata_manager,
+                    plugins,
+                )
+                .await?,
            ),
        })
    }
@@ -278,13 +287,20 @@ impl EngineInner {
        config: MitoConfig,
        log_store: Arc<S>,
        object_store_manager: ObjectStoreManagerRef,
+        schema_metadata_manager: SchemaMetadataManagerRef,
        plugins: Plugins,
    ) -> Result<EngineInner> {
        let config = Arc::new(config);
        let wal_raw_entry_reader = Arc::new(LogStoreRawEntryReader::new(log_store.clone()));
        Ok(EngineInner {
-            workers: WorkerGroup::start(config.clone(), log_store, object_store_manager, plugins)
-                .await?,
+            workers: WorkerGroup::start(
+                config.clone(),
+                log_store,
+                object_store_manager,
+                schema_metadata_manager,
+                plugins,
+            )
+            .await?,
            config,
            wal_raw_entry_reader,
        })
@@ -417,7 +433,7 @@ impl EngineInner {
        let version = region.version();
        // Get cache.
        let cache_manager = self.workers.cache_manager();
-        let scan_parallelism = ScanParallism {
+        let scan_parallelism = ScanParallelism {
            parallelism: self.config.scan_parallelism,
            channel_size: self.config.parallel_scan_channel_size,
        };
@@ -583,6 +599,7 @@ impl RegionEngine for MitoEngine {

 // Tests methods.
 #[cfg(any(test, feature = "test"))]
+#[allow(clippy::too_many_arguments)]
 impl MitoEngine {
    /// Returns a new [MitoEngine] for tests.
    pub async fn new_for_test<S: LogStore>(
@@ -593,6 +610,7 @@ impl MitoEngine {
        write_buffer_manager: Option<crate::flush::WriteBufferManagerRef>,
        listener: Option<crate::engine::listener::EventListenerRef>,
        time_provider: crate::time_provider::TimeProviderRef,
+        schema_metadata_manager: SchemaMetadataManagerRef,
    ) -> Result<MitoEngine> {
        config.sanitize(data_home)?;

@@ -606,6 +624,7 @@ impl MitoEngine {
                    object_store_manager,
                    write_buffer_manager,
                    listener,
+                    schema_metadata_manager,
                    time_provider,
                )
                .await?,
--- a/src/mito2/src/engine/alter_test.rs
+++ b/src/mito2/src/engine/alter_test.rs
@@ -22,7 +22,7 @@ use common_error::ext::ErrorExt;
 use common_error::status_code::StatusCode;
 use common_recordbatch::RecordBatches;
 use datatypes::prelude::ConcreteDataType;
-use datatypes::schema::ColumnSchema;
+use datatypes::schema::{ColumnSchema, FulltextAnalyzer, FulltextOptions};
 use store_api::metadata::ColumnMetadata;
 use store_api::region_engine::{RegionEngine, RegionRole};
 use store_api::region_request::{
@@ -31,7 +31,7 @@ use store_api::region_request::{
 use store_api::storage::{RegionId, ScanRequest};

 use crate::config::MitoConfig;
-use crate::engine::listener::AlterFlushListener;
+use crate::engine::listener::{AlterFlushListener, NotifyRegionChangeResultListener};
 use crate::engine::MitoEngine;
 use crate::test_util::{
    build_rows, build_rows_for_key, flush_region, put_rows, rows_schema, CreateRequestBuilder,
@@ -68,6 +68,36 @@ fn add_tag1() -> RegionAlterRequest {
    }
 }

+fn alter_column_fulltext_options() -> RegionAlterRequest {
+    RegionAlterRequest {
+        schema_version: 0,
+        kind: AlterKind::ChangeColumnFulltext {
+            column_name: "tag_0".to_string(),
+            options: FulltextOptions {
+                enable: true,
+                analyzer: FulltextAnalyzer::English,
+                case_sensitive: false,
+            },
+        },
+    }
+}
+
+fn check_region_version(
+    engine: &MitoEngine,
+    region_id: RegionId,
+    last_entry_id: u64,
+    committed_sequence: u64,
+    flushed_entry_id: u64,
+    flushed_sequence: u64,
+) {
+    let region = engine.get_region(region_id).unwrap();
+    let version_data = region.version_control.current();
+    assert_eq!(last_entry_id, version_data.last_entry_id);
+    assert_eq!(committed_sequence, version_data.committed_sequence);
+    assert_eq!(flushed_entry_id, version_data.version.flushed_entry_id);
+    assert_eq!(flushed_sequence, version_data.version.flushed_sequence);
+}
+
 #[tokio::test]
 async fn test_alter_region() {
    common_telemetry::init_default_ut_logging();
@@ -78,6 +108,16 @@ async fn test_alter_region() {
    let region_id = RegionId::new(1, 1);
    let request = CreateRequestBuilder::new().build();

+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let column_schemas = rows_schema(&request);
    let region_dir = request.region_dir.clone();
    engine
@@ -106,15 +146,7 @@ async fn test_alter_region() {
 |       | 2     | 2.0     | 1970-01-01T00:00:02 |
 +-------+-------+---------+---------------------+";
    scan_check_after_alter(&engine, region_id, expected).await;
-    let check_region = |engine: &MitoEngine| {
-        let region = engine.get_region(region_id).unwrap();
-        let version_data = region.version_control.current();
-        assert_eq!(1, version_data.last_entry_id);
-        assert_eq!(3, version_data.committed_sequence);
-        assert_eq!(1, version_data.version.flushed_entry_id);
-        assert_eq!(3, version_data.version.flushed_sequence);
-    };
-    check_region(&engine);
+    check_region_version(&engine, region_id, 1, 3, 1, 3);

    // Reopen region.
    let engine = env.reopen_engine(engine, MitoConfig::default()).await;
@@ -131,7 +163,7 @@ async fn test_alter_region() {
        .await
        .unwrap();
    scan_check_after_alter(&engine, region_id, expected).await;
-    check_region(&engine);
+    check_region_version(&engine, region_id, 1, 3, 1, 3);
 }

 /// Build rows with schema (string, f64, ts_millis, string).
@@ -167,10 +199,19 @@ fn build_rows_for_tags(
 async fn test_put_after_alter() {
    let mut env = TestEnv::new();
    let engine = env.create_engine(MitoConfig::default()).await;
-
    let region_id = RegionId::new(1, 1);
    let request = CreateRequestBuilder::new().build();

+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let mut column_schemas = rows_schema(&request);
    let region_dir = request.region_dir.clone();
    engine
@@ -266,6 +307,16 @@ async fn test_alter_region_retry() {
    let region_id = RegionId::new(1, 1);
    let request = CreateRequestBuilder::new().build();

+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let column_schemas = rows_schema(&request);
    engine
        .handle_request(region_id, RegionRequest::Create(request))
@@ -299,12 +350,7 @@ async fn test_alter_region_retry() {
 |       | a     | 1.0     | 1970-01-01T00:00:01 |
 +-------+-------+---------+---------------------+";
    scan_check_after_alter(&engine, region_id, expected).await;
-    let region = engine.get_region(region_id).unwrap();
-    let version_data = region.version_control.current();
-    assert_eq!(1, version_data.last_entry_id);
-    assert_eq!(2, version_data.committed_sequence);
-    assert_eq!(1, version_data.version.flushed_entry_id);
-    assert_eq!(2, version_data.version.flushed_sequence);
+    check_region_version(&engine, region_id, 1, 2, 1, 2);
 }

 #[tokio::test]
@@ -320,6 +366,16 @@ async fn test_alter_on_flushing() {
    let region_id = RegionId::new(1, 1);
    let request = CreateRequestBuilder::new().build();

+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let column_schemas = rows_schema(&request);
    engine
        .handle_request(region_id, RegionRequest::Create(request))
@@ -399,3 +455,187 @@ async fn test_alter_on_flushing() {
 +-------+-------+---------+---------------------+";
    assert_eq!(expected, batches.pretty_print().unwrap());
 }
+
+#[tokio::test]
+async fn test_alter_column_fulltext_options() {
+    common_telemetry::init_default_ut_logging();
+
+    let mut env = TestEnv::new();
+    let listener = Arc::new(AlterFlushListener::default());
+    let engine = env
+        .create_engine_with(MitoConfig::default(), None, Some(listener.clone()))
+        .await;
+
+    let region_id = RegionId::new(1, 1);
+    let request = CreateRequestBuilder::new().build();
+
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
+    let column_schemas = rows_schema(&request);
+    let region_dir = request.region_dir.clone();
+    engine
+        .handle_request(region_id, RegionRequest::Create(request))
+        .await
+        .unwrap();
+
+    let rows = Rows {
+        schema: column_schemas,
+        rows: build_rows(0, 3),
+    };
+    put_rows(&engine, region_id, rows).await;
+
+    // Spawns a task to flush the engine.
+    let engine_cloned = engine.clone();
+    let flush_job = tokio::spawn(async move {
+        flush_region(&engine_cloned, region_id, None).await;
+    });
+    // Waits for flush begin.
+    listener.wait_flush_begin().await;
+
+    // Consumes the notify permit in the listener.
+    listener.wait_request_begin().await;
+
+    // Submits an alter request to the region. The region should add the request
+    // to the pending ddl request list.
+    let request = alter_column_fulltext_options();
+    let engine_cloned = engine.clone();
+    let alter_job = tokio::spawn(async move {
+        engine_cloned
+            .handle_request(region_id, RegionRequest::Alter(request))
+            .await
+            .unwrap();
+    });
+    // Waits until the worker handles the alter request.
+    listener.wait_request_begin().await;
+
+    // Spawns two task to flush the engine. The flush scheduler should put them to the
+    // pending task list.
+    let engine_cloned = engine.clone();
+    let pending_flush_job = tokio::spawn(async move {
+        flush_region(&engine_cloned, region_id, None).await;
+    });
+    // Waits until the worker handles the flush request.
+    listener.wait_request_begin().await;
+
+    // Wake up flush.
+    listener.wake_flush();
+    // Wait for the flush job.
+    flush_job.await.unwrap();
+    // Wait for pending flush job.
+    pending_flush_job.await.unwrap();
+    // Wait for the write job.
+    alter_job.await.unwrap();
+
+    let expect_fulltext_options = FulltextOptions {
+        enable: true,
+        analyzer: FulltextAnalyzer::English,
+        case_sensitive: false,
+    };
+    let check_fulltext_options = |engine: &MitoEngine, expected: &FulltextOptions| {
+        let current_fulltext_options = engine
+            .get_region(region_id)
+            .unwrap()
+            .metadata()
+            .column_by_name("tag_0")
+            .unwrap()
+            .column_schema
+            .fulltext_options()
+            .unwrap()
+            .unwrap();
+        assert_eq!(*expected, current_fulltext_options);
+    };
+    check_fulltext_options(&engine, &expect_fulltext_options);
+    check_region_version(&engine, region_id, 1, 3, 1, 3);
+
+    // Reopen region.
+    let engine = env.reopen_engine(engine, MitoConfig::default()).await;
+    engine
+        .handle_request(
+            region_id,
+            RegionRequest::Open(RegionOpenRequest {
+                engine: String::new(),
+                region_dir,
+                options: HashMap::default(),
+                skip_wal_replay: false,
+            }),
+        )
+        .await
+        .unwrap();
+    check_fulltext_options(&engine, &expect_fulltext_options);
+    check_region_version(&engine, region_id, 1, 3, 1, 3);
+}
+
+#[tokio::test]
+async fn test_write_stall_on_altering() {
+    common_telemetry::init_default_ut_logging();
+
+    let mut env = TestEnv::new();
+    let listener = Arc::new(NotifyRegionChangeResultListener::default());
+    let engine = env
+        .create_engine_with(MitoConfig::default(), None, Some(listener.clone()))
+        .await;
+
+    let region_id = RegionId::new(1, 1);
+    let request = CreateRequestBuilder::new().build();
+
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
+    let column_schemas = rows_schema(&request);
+    engine
+        .handle_request(region_id, RegionRequest::Create(request))
+        .await
+        .unwrap();
+
+    let engine_cloned = engine.clone();
+    let alter_job = tokio::spawn(async move {
+        let request = add_tag1();
+        engine_cloned
+            .handle_request(region_id, RegionRequest::Alter(request))
+            .await
+            .unwrap();
+    });
+
+    let column_schemas_cloned = column_schemas.clone();
+    let engine_cloned = engine.clone();
+    let put_job = tokio::spawn(async move {
+        let rows = Rows {
+            schema: column_schemas_cloned,
+            rows: build_rows(0, 3),
+        };
+        put_rows(&engine_cloned, region_id, rows).await;
+    });
+
+    listener.wake_notify();
+    alter_job.await.unwrap();
+    put_job.await.unwrap();
+
+    let expected = "\
+-------+-------+---------+---------------------+
+| tag_1 | tag_0 | field_0 | ts                  |
+-------+-------+---------+---------------------+
+|       | 0     | 0.0     | 1970-01-01T00:00:00 |
+|       | 1     | 1.0     | 1970-01-01T00:00:01 |
+|       | 2     | 2.0     | 1970-01-01T00:00:02 |
+-------+-------+---------+---------------------+";
+    let request = ScanRequest::default();
+    let scanner = engine.scanner(region_id, request).unwrap();
+    let stream = scanner.scan().await.unwrap();
+    let batches = RecordBatches::try_collect(stream).await.unwrap();
+    assert_eq!(expected, batches.pretty_print().unwrap());
+}
--- a/src/mito2/src/engine/append_mode_test.rs
+++ b/src/mito2/src/engine/append_mode_test.rs
@@ -98,6 +98,16 @@ async fn test_append_mode_compaction() {
        .await;
    let region_id = RegionId::new(1, 1);

+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let request = CreateRequestBuilder::new()
        .insert_option("compaction.type", "twcs")
        .insert_option("compaction.twcs.max_active_window_runs", "2")
--- a/src/mito2/src/engine/compaction_test.rs
+++ b/src/mito2/src/engine/compaction_test.rs
@@ -112,6 +112,16 @@ async fn test_compaction_region() {
    let engine = env.create_engine(MitoConfig::default()).await;

    let region_id = RegionId::new(1, 1);
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let request = CreateRequestBuilder::new()
        .insert_option("compaction.type", "twcs")
        .insert_option("compaction.twcs.max_active_window_runs", "1")
@@ -171,8 +181,18 @@ async fn test_compaction_region_with_overlapping() {
    common_telemetry::init_default_ut_logging();
    let mut env = TestEnv::new();
    let engine = env.create_engine(MitoConfig::default()).await;
-
    let region_id = RegionId::new(1, 1);
+
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let request = CreateRequestBuilder::new()
        .insert_option("compaction.type", "twcs")
        .insert_option("compaction.twcs.max_active_window_runs", "2")
@@ -217,6 +237,17 @@ async fn test_compaction_region_with_overlapping_delete_all() {
    let engine = env.create_engine(MitoConfig::default()).await;

    let region_id = RegionId::new(1, 1);
+
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let request = CreateRequestBuilder::new()
        .insert_option("compaction.type", "twcs")
        .insert_option("compaction.twcs.max_active_window_runs", "2")
@@ -281,6 +312,16 @@ async fn test_readonly_during_compaction() {
        .await;

    let region_id = RegionId::new(1, 1);
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    let request = CreateRequestBuilder::new()
        .insert_option("compaction.type", "twcs")
        .insert_option("compaction.twcs.max_active_window_runs", "1")
--- a/src/mito2/src/engine/drop_test.rs
+++ b/src/mito2/src/engine/drop_test.rs
@@ -16,6 +16,7 @@ use std::sync::Arc;
 use std::time::Duration;

 use api::v1::Rows;
+use common_meta::key::SchemaMetadataManager;
 use object_store::util::join_path;
 use store_api::region_engine::RegionEngine;
 use store_api::region_request::{RegionDropRequest, RegionRequest};
@@ -40,6 +41,17 @@ async fn test_engine_drop_region() {
        .await;

    let region_id = RegionId::new(1, 1);
+
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
+
    // It's okay to drop a region doesn't exist.
    engine
        .handle_request(region_id, RegionRequest::Drop(RegionDropRequest {}))
@@ -87,7 +99,12 @@ async fn test_engine_drop_region() {
 #[tokio::test]
 async fn test_engine_drop_region_for_custom_store() {
    common_telemetry::init_default_ut_logging();
-    async fn setup(engine: &MitoEngine, region_id: RegionId, storage_name: &str) {
+    async fn setup(
+        engine: &MitoEngine,
+        schema_metadata_manager: &SchemaMetadataManager,
+        region_id: RegionId,
+        storage_name: &str,
+    ) {
        let request = CreateRequestBuilder::new()
            .insert_option("storage", storage_name)
            .region_dir(storage_name)
@@ -97,6 +114,18 @@ async fn test_engine_drop_region_for_custom_store() {
            .handle_request(region_id, RegionRequest::Create(request))
            .await
            .unwrap();
+
+        let table_id = format!("test_table_{}", region_id.table_id());
+        schema_metadata_manager
+            .register_region_table_info(
+                region_id.table_id(),
+                &table_id,
+                "test_catalog",
+                "test_schema",
+                None,
+            )
+            .await;
+
        let rows = Rows {
            schema: column_schema.clone(),
            rows: build_rows_for_key("a", 0, 2, 0),
@@ -114,12 +143,19 @@ async fn test_engine_drop_region_for_custom_store() {
            &["Gcs"],
        )
        .await;
+    let schema_metadata_manager = env.get_schema_metadata_manager();
    let object_store_manager = env.get_object_store_manager().unwrap();

    let global_region_id = RegionId::new(1, 1);
-    setup(&engine, global_region_id, "default").await;
+    setup(
+        &engine,
+        &schema_metadata_manager,
+        global_region_id,
+        "default",
+    )
+    .await;
    let custom_region_id = RegionId::new(2, 1);
-    setup(&engine, custom_region_id, "Gcs").await;
+    setup(&engine, &schema_metadata_manager, custom_region_id, "Gcs").await;

    let global_region = engine.get_region(global_region_id).unwrap();
    let global_region_dir = global_region.access_layer.region_dir().to_string();
--- a/src/mito2/src/engine/edit_region_test.rs
+++ b/src/mito2/src/engine/edit_region_test.rs
@@ -64,6 +64,16 @@ async fn test_edit_region_schedule_compaction() {
        .await;

    let region_id = RegionId::new(1, 1);
+
+    env.get_schema_metadata_manager()
+        .register_region_table_info(
+            region_id.table_id(),
+            "test_table",
+            "test_catalog",
+            "test_schema",
+            None,
+        )
+        .await;
    engine
        .handle_request(
            region_id,
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
evenyag	ce0031067a	docs: how to get jeprof from target dir	2024-11-14 14:21:48 +08:00
evenyag	c98db4f5ff	build: add jeprof to linux image	2024-11-14 14:21:36 +08:00
evenyag	975f19f452	build: add jeprof to image	2024-11-13 23:04:43 +08:00
dennis zhuang	d7c3c8e124	fix: physical table statistics info (#4975 ) * fix: physical table statistics info * refactor: is_physical_table * fix: remove file	2024-11-13 14:29:44 +00:00
Yohan Wal	f4b9eac465	build(deps): switch to upstream jsonb (#4986 ) chore: jsonb	2024-11-13 11:15:37 +00:00
Yohan Wal	aa6c2de42a	refactor: use UNSET instead of enable (#4983 ) * refactor: use UNSERT instead of enable * fix: test * chore: comment	2024-11-13 08:28:20 +00:00
discord9	175fddb3b5	fix: alter table add column id alloc mismatch (#4972 ) * fix: alter table add column id alloc mismatch * chore: remove debug code * chore: typos * chore: error variant * chore: more checks for invariant * refactor: better check&explain * fix: exist column metadata correct * chore: remove unused error variant	2024-11-13 07:02:35 +00:00
Lei, HUANG	6afc4e778a	refactor(mito): tidy memtable stats (#4982 ) * wip: share same WriteMetrics struct between different memtable implementations * refactor: extract function to update memtable timestamp range	2024-11-13 06:57:27 +00:00
Lanqing Yang	3bbcde8e58	feat: support alter twcs compaction options (#4965 ) support alter twcs compression options	2024-11-13 06:27:32 +00:00
Weny Xu	3bf9981aab	refactor: support distinct JSON format and improve type conversions (#4979 ) refactor: support distinct JSON format	2024-11-13 03:03:51 +00:00
Weny Xu	c47ad548a4	feat: refine region state checks and handle stalled requests (#4971 ) * feat: refine region state checks and handle stalled requests * test: add tests * chore: apply suggestions from CR * chore: apply suggestions from CR * chore: add comments	2024-11-13 02:51:27 +00:00
Lin Yihai	0b6d78a527	refactor: consolidate `DatanodeClientOptions` (#4966 ) refactor!: consolidate `DatanodeClientOptions`	2024-11-12 09:57:41 +00:00
Zhenchi	d616bd92ef	feat: introduce vector type (#4964 ) * feat: introduce vector type Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * test: fix prepared stmt Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * test: add grpc test Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * test: parse vector value Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * test: column to row Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * test: sqlness Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * fix: merge issue Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * refactor: add check for bytes size Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * Update tests/cases/standalone/common/types/vector/vector.sql Co-authored-by: Ruihang Xia <waynestxia@gmail.com> * chore: update proto Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * chore: simplify cargo Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * chore: address comment Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> --------- Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> Co-authored-by: Ruihang Xia <waynestxia@gmail.com>	2024-11-12 08:28:44 +00:00
Yohan Wal	84aa5b7b22	feat: alter fulltext options (#4952 ) * feat(WIP): alter fulltext index Co-Authored-By: irenjj <renj.jiang@gmail.com> * feat: alter column fulltext option Co-Authored-By: irenjj <renj.jiang@gmail.com> * chore: fmt * test: add unit and integration tests Co-Authored-By: irenjj <renj.jiang@gmail.com> * test: update sqlness test * chore: new line * chore: lock file update * chore: apply review comments * test: update sqlness test * test: update sqlness test * fix: convert * chore: apply review comments * fix: toml fmt * fix: tests * test: add test for mito * chore: error message * fix: test * fix: test * fix: wrong comment * chore: change proto rev * chore: apply review comments * chore: apply review comments * chore: fmt --------- Co-authored-by: irenjj <renj.jiang@gmail.com>	2024-11-12 03:04:04 +00:00
Zhenchi	cbf21e53a9	feat(puffin): apply range reader (#4928 ) * wip Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * feat(puffin): apply range reader Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * refactor: read_vec reduce iteration Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * refactor: simplify rangereader for vec<u8> Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * test: add unit test Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * fix: toml format Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> --------- Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>	2024-11-12 02:36:38 +00:00
Zhenchi	6248a6ccf5	feat(index): support SQL to specify inverted index columns (#4929 ) * feat(index): support building inverted index for the field column Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * feat(index): support SQL to specify inverted index columns Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * test: fix sqlness Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * fix: consider compatibility Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * polish Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * compatibility Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * fix Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * fix: ignore case Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * refactor: reduce dup Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * fix: clippy Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> --------- Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>	2024-11-11 08:06:23 +00:00
localhost	0e0c4faf0d	fix(otlp): replace otlp trace attr type from string to jsonb (#4918 ) * chore: minor update * chore: replace otlp trace attr type from string to jsonb * chore: add new util file and remove useless code * chore: add license header * chore: remove unused error * chore: adjust otlp traces column order * chore: update test * chore: minor fix --------- Co-authored-by: shuiyisong <xixing.sys@gmail.com>	2024-11-08 06:34:49 +00:00
Kaifeng Zheng	1a02fc31c2	fix: json_path_exists null results (#4881 ) * fix: result of nulls * update test result * fix null behaviors, add null tests * update NULL tests * error handler when parsing json_path * change the logic to: items' datatype in the input arrays are all the same. * remove a comment * refactor: better logic * drop unnecessary err check * added an error test case	2024-11-08 03:01:45 +00:00
Ruihang Xia	8efbafa538	feat: support filter with windowed sort (#4960 ) * feat: support windowed sort with where condition Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * fix split logic Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * modify fuzz test to reflect logic change Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * feat: handle sort that wont preserving partition Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * clean up Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * fix typo Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * fix test case and add more cases Signed-off-by: Ruihang Xia <waynestxia@gmail.com> --------- Signed-off-by: Ruihang Xia <waynestxia@gmail.com>	2024-11-08 02:49:36 +00:00
Ruihang Xia	fcd0ceea94	fix: column already exists (#4961 ) * fix: merge fetched logical metadata with existing cache Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * fix log acquire Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * Update src/metric-engine/src/engine/region_metadata.rs Co-authored-by: Yingwen <realevenyag@gmail.com> --------- Signed-off-by: Ruihang Xia <waynestxia@gmail.com> Co-authored-by: Yingwen <realevenyag@gmail.com>	2024-11-07 13:25:05 +00:00
jeremyhi	22f31f5929	chore: paginated query region stats (#4942 )	2024-11-07 03:01:12 +00:00
Lei, HUANG	5d20acca44	fix: round euclidean result in sqlness (#4956 )	2024-11-07 02:29:49 +00:00
Yingwen	e3733344fe	fix: do not pick compacting/expired files (#4955 )	2024-11-06 21:38:26 +00:00
dennis zhuang	305767e226	fix: bugs introduced by alter table options (#4953 ) * fix: ChangeTableOptions display * fix: partition number disappear after altering table options * Update src/table/src/metadata.rs Co-authored-by: Ruihang Xia <waynestxia@gmail.com> --------- Co-authored-by: Lei, HUANG <6406592+v0y4g3r@users.noreply.github.com> Co-authored-by: Ruihang Xia <waynestxia@gmail.com>	2024-11-06 19:56:49 +00:00
dennis zhuang	22a662f6bc	docs: add TOC to readme (#4949 ) * docs: add TOC to readme * chore: remove some links * chore: tweak TOC * chore: tweak TOC	2024-11-06 13:27:00 +00:00
Lin Yihai	1431393fc8	fix: the region_stats API will return an error in instance test (#4951 )	2024-11-06 13:15:48 +00:00
localhost	dfe8cf25f9	chore: add json path for pipeline (#4925 ) * chore: add json path for pipeline * chore: change jsonpath lib verion * chore: remove useless doc * chore: fix json path test * chore: fix pipeline json path test	2024-11-06 11:44:59 +00:00
Weny Xu	cccd25ddbb	chore: fix typos in change log level doc (#4948 ) chore: fix typos in change log level	2024-11-06 07:49:00 +00:00
discord9	ac387bd2af	fix: pprof (#4938 )	2024-11-05 09:22:29 +00:00
LFC	2e9737c01d	refactor: pass `LogicalPlan` to promql execution interceptor (#4937 )	2024-11-05 08:49:05 +00:00
Ning Sun	a8b426aebe	feat: add more geo functions (#4888 ) * chore: add type conversion for array types * feat: add h3_cells_contains * refactor: resolve lint issues * feat: add sphere distance function * feat: euclidean distance between h3 centroids * test: round float number * feat: add more geospatial functions * test: add tests for geometry functions * refactor: move wkt function to dedicated module * feat: add st_area * refactor: only allow sphere distance between points	2024-11-05 03:44:25 +00:00
jeremyhi	f3509fa312	chore: minor refactor for weighted choose (#4917 ) * chore: minor refactor for weighted choose * chore: by comment, remove the fast path of choose_multiple	2024-11-05 02:55:11 +00:00
Lei, HUANG	3dcd6b8e51	fix: database base ttl (#4926 ) * main: Add common-meta dependency and implement SchemaMetadataManager - Introduce `common-meta` as a new dependency in `mito2`. - Implement `SchemaMetadataManager` for managing schema-level metadata. - Update `DatanodeBuilder` and `MitoEngine` to pass `KvBackendRef` for schema metadata management. - Add `SchemaMetadataManager` to `RegionWorkerLoop` for compaction handling. - Include `SchemaNameKey` usage in compaction-related code. - Add `database_metadata_manager` module with `SchemaMetadataManager` struct and associated logic. * fix/database-base-ttl: Refactor metadata management and update compaction logic - Remove `database_metadata_manager` and introduce `schema_metadata_manager` - Update compaction logic to handle TTL based on schema metadata - Adjust tests to use `schema_metadata_manager` for setting up schema options - Fix engine creation in tests to pass `kv_backend` explicitly - Remove unused imports and apply minor code cleanups * fix/database-base-ttl: Extend CREATE TABLE LIKE to inherit schema options - Implement inheritance of database level options for CREATE TABLE LIKE - Add schema options to SHOW CREATE TABLE output - Refactor create_table_stmt to include schema_options in SQL generation - Update error handling to include TableMetadataManagerSnafu * fix/database-base-ttl: Refactor error handling and remove schema dependency in table creation - Replace expect with the ? operator for error handling in open_compaction_region - Simplify create_logical_tables by removing catalog and schema name parameters - Remove unnecessary schema retrieval and merging of schema options in create_table_info - Clean up unused imports and redundant code * fix/database-base-ttl: Refactor error handling and update documentation comments - Update comment to reflect retrieval of schema options instead of metadata - Introduce new error type `GetSchemaMetadataSnafu` for schema metadata retrieval failures - Implement error handling for schema metadata retrieval in `find_ttl` function * fix: toml * fix/database-base-ttl: Refactor SchemaMetadataManager and adjust Cargo.toml dependencies - Remove unused imports in schema_metadata_manager.rs - Add conditional compilation for SchemaMetadataManager::new - Update Cargo.toml to remove "testing" feature from common-meta dependency in main section and add it to dev-dependencies * fix/database-base-ttl: Fix typos in comments and function names across multiple modules - Correct spelling of 'parallelism' in region_server, engine, and scan_region modules - Amend typo in TODO comment from 'persisent' to 'persistent' in server module - Update incorrect test query from 'versiona' to 'version' in federated module tests * fix/database-base-ttl: Add schema existence check in StatementExecutor for CREATE TABLE operation * fix/database-base-ttl: Add warning log for failed TTL retrieval in compaction region open function * fix/database-base-ttl: Refactor to use SchemaMetadataManagerRef in Datanode and MitoEngine - Replace KvBackendRef with SchemaMetadataManagerRef across various components. - Update DatanodeBuilder and MitoEngine to pass SchemaMetadataManagerRef instead of KvBackendRef. - Adjust test cases to use get_schema_metadata_manager method for consistency.	2024-11-05 02:51:32 +00:00
Weny Xu	f221ee30fd	fix: violations of `elided_named_lifetimes` (#4936 )	2024-11-04 10:52:39 +00:00
Yohan Wal	fb822987a9	refactor: refactor alter parser (#4933 ) refactor: alter parser	2024-11-04 09:00:30 +00:00
Weny Xu	4ab6dc2825	feat: support to insert json data via grpc protocol (#4908 ) * feat: support to insert json data via grpc protocol * chore: handle error * feat: introduce `prepare_rows` * chore: fmt toml * test: add row deletion test * test: fix unit test * chore: remove log * chore: apply suggestions from CR	2024-11-04 08:55:47 +00:00
dennis zhuang	191755fc42	fix: data_length, index_length, table_rows in tables (#4927 ) * fix: data_length, index_length, table_rows in tables * feat: table stats only works for mito engine currently * fix: tests * fix: typo * chore: log error when region_stats fails	2024-11-04 07:44:13 +00:00
Yohan Wal	1676d02149	fix: panic when jsonb corrupted (#4919 ) * refactor: json type update * test: update test * fix: convert when needed * revert: leave sqlness tests unchanged * fix: fmt * refactor: just refactor * Apply suggestions from code review Co-authored-by: Weny Xu <wenymedia@gmail.com> * refactor: parse jsonb first * test: add bad cases * Update src/datatypes/src/vectors/binary.rs Co-authored-by: Weny Xu <wenymedia@gmail.com> * fix: fmt * fix: fix clippy/check * fix: corrupted jsonb panic * chore(deps): change to rev --------- Co-authored-by: Weny Xu <wenymedia@gmail.com>	2024-11-04 06:55:14 +00:00
dennis zhuang	edc49623de	chore: update default cache size to 1Gib (#4923 ) * chore: update default cache size to 1Gib for object storage read/write cache * feat: update docs * fix: test	2024-11-04 03:53:17 +00:00
jeremyhi	9405d1c578	feat: heartbeat_flush_threshold option (#4924 ) * feat: heartbeat_flush_threshold * chore: rename to flush_stats_factor * Update src/meta-srv/src/handler/collect_stats_handler.rs	2024-11-04 03:34:50 +00:00
dennis zhuang	7a4276c24a	fix: typo (#4931 ) fix/database-base-ttl: Fix typos in comments and function names across multiple modules - Correct spelling of 'parallelism' in region_server, engine, and scan_region modules - Amend typo in TODO comment from 'persisent' to 'persistent' in server module - Update incorrect test query from 'versiona' to 'version' in federated module tests Co-authored-by: Lei, HUANG <mrsatangel@gmail.com>	2024-11-04 01:56:11 +00:00