diff --git a/.github/release-drafter.yml b/.github/release-drafter.yml
deleted file mode 100644
index f6ea319..0000000
--- a/.github/release-drafter.yml
+++ /dev/null
@@ -1,7 +0,0 @@
-# https://github.com/release-drafter/release-drafter#example
-name-template: 'resp-benchmark-v$RESOLVED_VERSION'
-tag-template: 'v$RESOLVED_VERSION'
-template: |
-  ## Changes
-
-  $CHANGES
diff --git a/.github/workflows/CI.yml b/.github/workflows/CI.yml
new file mode 100644
index 0000000..ce2bc11
--- /dev/null
+++ b/.github/workflows/CI.yml
@@ -0,0 +1,165 @@
+name: CI
+
+on:
+  push:
+    branches:
+      - main
+      - master
+    tags:
+      - '*'
+  pull_request:
+  workflow_dispatch:
+
+
+permissions:
+  contents: read
+
+jobs:
+  linux:
+    runs-on: ${{ matrix.platform.runner }}
+    strategy:
+      matrix:
+        platform:
+          - runner: ubuntu-latest
+            target: x86_64
+          - runner: ubuntu-latest
+            target: x86
+          - runner: ubuntu-latest
+            target: aarch64
+          - runner: ubuntu-latest
+            target: armv7
+          - runner: ubuntu-latest
+            target: s390x
+          - runner: ubuntu-latest
+            target: ppc64le
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: 3.x
+      - name: Build wheels
+        uses: PyO3/maturin-action@v1
+        with:
+          target: ${{ matrix.platform.target }}
+          args: --release --out dist --find-interpreter
+          sccache: 'true'
+          manylinux: auto
+      - name: Upload wheels
+        uses: actions/upload-artifact@v4
+        with:
+          name: wheels-linux-${{ matrix.platform.target }}
+          path: dist
+
+  musllinux:
+    runs-on: ${{ matrix.platform.runner }}
+    strategy:
+      matrix:
+        platform:
+          - runner: ubuntu-latest
+            target: x86_64
+          - runner: ubuntu-latest
+            target: x86
+          - runner: ubuntu-latest
+            target: aarch64
+          - runner: ubuntu-latest
+            target: armv7
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: 3.x
+      - name: Build wheels
+        uses: PyO3/maturin-action@v1
+        with:
+          target: ${{ matrix.platform.target }}
+          args: --release --out dist --find-interpreter
+          sccache: 'true'
+          manylinux: musllinux_1_2
+      - name: Upload wheels
+        uses: actions/upload-artifact@v4
+        with:
+          name: wheels-musllinux-${{ matrix.platform.target }}
+          path: dist
+
+  windows:
+    runs-on: ${{ matrix.platform.runner }}
+    strategy:
+      matrix:
+        platform:
+          - runner: windows-latest
+            target: x64
+          - runner: windows-latest
+            target: x86
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: 3.x
+          architecture: ${{ matrix.platform.target }}
+      - name: Build wheels
+        uses: PyO3/maturin-action@v1
+        with:
+          target: ${{ matrix.platform.target }}
+          args: --release --out dist --find-interpreter
+          sccache: 'true'
+      - name: Upload wheels
+        uses: actions/upload-artifact@v4
+        with:
+          name: wheels-windows-${{ matrix.platform.target }}
+          path: dist
+
+  macos:
+    runs-on: ${{ matrix.platform.runner }}
+    strategy:
+      matrix:
+        platform:
+          - runner: macos-12
+            target: x86_64
+          - runner: macos-14
+            target: aarch64
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: 3.x
+      - name: Build wheels
+        uses: PyO3/maturin-action@v1
+        with:
+          target: ${{ matrix.platform.target }}
+          args: --release --out dist --find-interpreter
+          sccache: 'true'
+      - name: Upload wheels
+        uses: actions/upload-artifact@v4
+        with:
+          name: wheels-macos-${{ matrix.platform.target }}
+          path: dist
+
+  sdist:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Build sdist
+        uses: PyO3/maturin-action@v1
+        with:
+          command: sdist
+          args: --out dist
+      - name: Upload sdist
+        uses: actions/upload-artifact@v4
+        with:
+          name: wheels-sdist
+          path: dist
+
+  release:
+    name: Release
+    runs-on: ubuntu-latest
+    if: "startsWith(github.ref, 'refs/tags/')"
+    needs: [linux, musllinux, windows, macos, sdist]
+    steps:
+      - uses: actions/download-artifact@v4
+      - name: Publish to PyPI
+        uses: PyO3/maturin-action@v1
+        env:
+          MATURIN_PYPI_TOKEN: ${{ secrets.PYPI_API_TOKEN }}
+        with:
+          command: upload
+          args: --non-interactive --skip-existing wheels-*/*
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index 33338e1..5baad95 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -1,122 +1,23 @@
-name: release
+name: Release
 
-# Controls when the workflow will run
 on:
-  # Allows you to run this workflow manually from the Actions tab
   workflow_dispatch:
 
 jobs:
-  draft:
-    name: draft
+  release:
     runs-on: ubuntu-latest
     steps:
-      - name: draft release
-        id: "draft"
-        uses: release-drafter/release-drafter@v5
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-    outputs:
-      upload_url: ${{ steps.draft.outputs.upload_url }}
-
-  build-and-upload:
-    needs: draft
-    name: build - ${{ matrix.platform.release_for }}
-    strategy:
-      matrix:
-        platform:
-          - release_for: linux-amd64
-            os: ubuntu-20.04
-            target: x86_64-unknown-linux-gnu
-            bin: resp-benchmark
-            name: resp-benchmark-linux-amd64.tar.gz
-            command: build
-
-          - release_for: linux-arm64
-            os: ubuntu-20.04
-            target: aarch64-unknown-linux-gnu
-            bin: resp-benchmark
-            name: resp-benchmark-linux-arm64.tar.gz
-            command: build
-
-          - release_for: windows-amd64
-            os: windows-latest
-            target: x86_64-pc-windows-msvc
-            bin: resp-benchmark.exe
-            name: resp-benchmark-windows-amd64.zip
-            command: build
-
-          - release_for: windows-arm64
-            os: windows-latest
-            target: aarch64-pc-windows-msvc
-            bin: resp-benchmark.exe
-            name: resp-benchmark-windows-arm64.zip
-            command: build
-
-          - release_for: macos-amd64
-            os: macos-latest
-            target: x86_64-apple-darwin
-            bin: resp-benchmark
-            name: resp-benchmark-macos-amd64.tar.gz
-            command: build
-
-          - release_for: macos-arm64
-            os: macos-latest
-            target: aarch64-apple-darwin
-            bin: resp-benchmark
-            name: resp-benchmark-macos-arm64.tar.gz
-            command: build
+      - uses: actions/checkout@v2
 
-    runs-on: ${{ matrix.platform.os }}
-    steps:
-      - name: checkout
-        uses: actions/checkout@v3
-
-      - name: cache cargo registry
-        uses: actions/cache@v3
+      - name: Set up Rust
+        uses: actions/setup-rust@v1
         with:
-          path: ~/.cargo/registry
-          key: ${{ runner.os }}-cargo-registry-${{ hashFiles('**/Cargo.lock') }}
-          restore-keys: |
-            ${{ runner.os }}-cargo-registry-
+          rust-version: stable
 
-      - name: cache cargo index
-        uses: actions/cache@v3
-        with:
-          path: ~/.cargo/git
-          key: ${{ runner.os }}-cargo-index-${{ hashFiles('**/Cargo.lock') }}
-          restore-keys: |
-            ${{ runner.os }}-cargo-index-
+      - name: Install cargo-release
+        run: cargo install cargo-release
 
-      - name: cache cargo build
-        uses: actions/cache@v3
-        with:
-          path: target
-          key: ${{ runner.os }}-cargo-build-${{ hashFiles('**/Cargo.lock') }}
-          restore-keys: |
-            ${{ runner.os }}-cargo-build-
-
-      - name: build binary
-        uses: houseabsolute/actions-rust-cross@v0
-        with:
-          command: ${{ matrix.platform.command }}
-          target: ${{ matrix.platform.target }}
-          args: "--locked --release"
-          strip: true
-      - name: Package as archive
-        shell: bash
-        run: |
-          cd target/${{ matrix.platform.target }}/release
-          if [[ "${{ matrix.platform.os }}" == "windows-latest" ]]; then
-            7z a ${{ matrix.platform.name }} ${{ matrix.platform.bin }} ../../../workloads
-          else
-            tar czvf ${{ matrix.platform.name }} ${{ matrix.platform.bin }} ../../../workloads
-          fi
-      - name: upload
-        uses: actions/upload-release-asset@v1
-        with:
-          upload_url: ${{ needs.draft.outputs.upload_url }}
-          asset_path: ./target/${{ matrix.platform.target }}/release/${{ matrix.platform.name }}
-          asset_name: ${{ matrix.platform.name }}
-          asset_content_type: ${{ matrix.platform.os == 'windows-latest' && 'application/zip' || 'application/gzip' }}
+      - name: Run cargo release
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: cargo release patch --no-publish
diff --git a/.gitignore b/.gitignore
index 96ef862..c8f0442 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,72 @@
-target/
+/target
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+.pytest_cache/
+*.py[cod]
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+.venv/
+env/
+bin/
+build/
+develop-eggs/
+dist/
+eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+include/
+man/
+venv/
+*.egg-info/
+.installed.cfg
+*.egg
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+pip-selfcheck.json
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.cache
+nosetests.xml
+coverage.xml
+
+# Translations
+*.mo
+
+# Mr Developer
+.mr.developer.cfg
+.project
+.pydevproject
+
+# Rope
+.ropeproject
+
+# Django stuff:
+*.log
+*.pot
+
+.DS_Store
+
+# Sphinx documentation
+docs/_build/
+
+# PyCharm
 .idea/
+
+# VSCode
+.vscode/
+
+# Pyenv
+.python-version
diff --git a/Cargo.lock b/Cargo.lock
index f64de0a..1352bd8 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -18,93 +18,20 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f26201604c87b1e01bd3d98f8d5d9a8fcbb815e8cedb41ffccbeb4bf593a35fe"
 
 [[package]]
-name = "aho-corasick"
-version = "1.1.3"
+name = "arc-swap"
+version = "1.7.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8e60d3430d3a69478ad0993f19238d2df97c507009a52b3c10addcd7f6bcb916"
-dependencies = [
- "memchr",
-]
-
-[[package]]
-name = "android-tzdata"
-version = "0.1.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e999941b234f3131b00bc13c22d06e8c5ff726d1b6318ac7eb276997bbb4fef0"
-
-[[package]]
-name = "android_system_properties"
-version = "0.1.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "819e7219dbd41043ac279b19830f2efc897156490d7fd6ea916720117ee66311"
-dependencies = [
- "libc",
-]
-
-[[package]]
-name = "anstream"
-version = "0.6.14"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "418c75fa768af9c03be99d17643f93f79bbba589895012a80e3452a19ddda15b"
-dependencies = [
- "anstyle",
- "anstyle-parse",
- "anstyle-query",
- "anstyle-wincon",
- "colorchoice",
- "is_terminal_polyfill",
- "utf8parse",
-]
-
-[[package]]
-name = "anstyle"
-version = "1.0.7"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "038dfcf04a5feb68e9c60b21c9625a54c2c0616e79b72b0fd87075a056ae1d1b"
-
-[[package]]
-name = "anstyle-parse"
-version = "0.2.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c03a11a9034d92058ceb6ee011ce58af4a9bf61491aa7e1e59ecd24bd40d22d4"
-dependencies = [
- "utf8parse",
-]
-
-[[package]]
-name = "anstyle-query"
-version = "1.1.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ad186efb764318d35165f1758e7dcef3b10628e26d41a44bc5550652e6804391"
-dependencies = [
- "windows-sys 0.52.0",
-]
-
-[[package]]
-name = "anstyle-wincon"
-version = "3.0.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "61a38449feb7068f52bb06c12759005cf459ee52bb4adc1d5a7c4322d716fb19"
-dependencies = [
- "anstyle",
- "windows-sys 0.52.0",
-]
-
-[[package]]
-name = "anyhow"
-version = "1.0.86"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b3d1d046238990b9cf5bcde22a3fb3584ee5cf65fb2765f454ed428c7a0063da"
+checksum = "69f7f8c3906b62b754cd5326047894316021dcfe5a194c8ea52bdd94934a3457"
 
 [[package]]
 name = "async-trait"
-version = "0.1.80"
+version = "0.1.81"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c6fa2087f2753a7da8cc1c0dbfcf89579dd57458e36769de5ac750b4671737ca"
+checksum = "6e0c28dcc82d7c8ead5cb13beb15405b57b8546e93215673ff8ca0349a028107"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.68",
+ "syn 2.0.72",
 ]
 
 [[package]]
@@ -134,23 +61,11 @@ dependencies = [
  "rustc-demangle",
 ]
 
-[[package]]
-name = "base64"
-version = "0.22.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "72b3254f16251a8381aa12e40e3c4d2f0199f8c6508fbecb9d91f575e0fbb8c6"
-
 [[package]]
 name = "bitflags"
-version = "2.5.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cf4b9d6a944f767f8e5e0db018570623c85f3d925ac718db4e06d0187adb21c1"
-
-[[package]]
-name = "bumpalo"
-version = "3.16.0"
+version = "2.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "79296716171880943b8470b5f8d03aa55eb2e645a4874bdbb28adb49162e012c"
+checksum = "b048fb63fd8b5923fc5aa7b340d8e156aec7ec02f0c78fa8a6ddc2613f6f71de"
 
 [[package]]
 name = "byteorder"
@@ -160,15 +75,15 @@ checksum = "1fd0f2584146f6f2ef48085050886acf353beff7305ebd1ae69500e27c67f64b"
 
 [[package]]
 name = "bytes"
-version = "1.6.0"
+version = "1.7.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "514de17de45fdb8dc022b1a7975556c53c86f9f0aa5f534b98977b171857c2c9"
+checksum = "8318a53db07bb3f8dca91a600466bdb3f2eaadeedfdbcf02e1accbad9271ba50"
 
 [[package]]
 name = "cc"
-version = "1.0.100"
+version = "1.1.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c891175c3fb232128f48de6590095e59198bbeb8620c310be349bfc3afd12c7b"
+checksum = "504bdec147f2cc13c8b57ed9401fd8a147cc66b67ad5cb241394244f2c947549"
 
 [[package]]
 name = "cfg-if"
@@ -177,64 +92,10 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
 
 [[package]]
-name = "chrono"
-version = "0.4.38"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a21f936df1771bf62b77f047b726c4625ff2e8aa607c01ec06e5a05bd8463401"
-dependencies = [
- "android-tzdata",
- "iana-time-zone",
- "js-sys",
- "num-traits",
- "wasm-bindgen",
- "windows-targets 0.52.5",
-]
-
-[[package]]
-name = "clap"
-version = "4.5.7"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5db83dced34638ad474f39f250d7fea9598bdd239eaced1bdf45d597da0f433f"
-dependencies = [
- "clap_builder",
- "clap_derive",
-]
-
-[[package]]
-name = "clap_builder"
-version = "4.5.7"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f7e204572485eb3fbf28f871612191521df159bc3e15a9f5064c66dba3a8c05f"
-dependencies = [
- "anstream",
- "anstyle",
- "clap_lex",
- "strsim",
-]
-
-[[package]]
-name = "clap_derive"
-version = "4.5.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c780290ccf4fb26629baa7a1081e68ced113f1d3ec302fa5948f1c381ebf06c6"
-dependencies = [
- "heck",
- "proc-macro2",
- "quote",
- "syn 2.0.68",
-]
-
-[[package]]
-name = "clap_lex"
-version = "0.7.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4b82cf0babdbd58558212896d1a4272303a57bdb245c2bf1147185fb45640e70"
-
-[[package]]
-name = "colorchoice"
-version = "1.0.1"
+name = "cfg_aliases"
+version = "0.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0b6a852b24ab71dffc585bcb46eaf7959d175cb865a7152e35b348d1b2960422"
+checksum = "fd16c4719339c4530435d38e511904438d07cce7950afa3718a84ac36c10e89e"
 
 [[package]]
 name = "colored"
@@ -260,22 +121,6 @@ dependencies = [
  "tokio-util",
 ]
 
-[[package]]
-name = "core-foundation"
-version = "0.9.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "91e195e091a93c46f7102ec7818a2aa394e1e1771c3ab4825963fa03e45afb8f"
-dependencies = [
- "core-foundation-sys",
- "libc",
-]
-
-[[package]]
-name = "core-foundation-sys"
-version = "0.8.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "06ea2b9bc92be3c2baa9334a323ebca2d6f074ff852cd1d7b11064035cd3868f"
-
 [[package]]
 name = "core_affinity"
 version = "0.8.1"
@@ -294,26 +139,15 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "338089f42c427b86394a5ee60ff321da23a5c89c9d89514c829687b26359fcff"
 
 [[package]]
-name = "crc32fast"
-version = "1.4.2"
+name = "ctrlc"
+version = "3.4.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a97769d94ddab943e4510d138150169a2758b5ef3eb191a9ee688de3e23ef7b3"
+checksum = "672465ae37dc1bc6380a6547a8883d5dd397b0f1faaad4f265726cc7042a5345"
 dependencies = [
- "cfg-if",
+ "nix",
+ "windows-sys 0.52.0",
 ]
 
-[[package]]
-name = "crossbeam-utils"
-version = "0.8.20"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "22ec99545bb0ed0ea7bb9b8e1e9122ea386ff8a48c0922e43f36d45ab09e0e80"
-
-[[package]]
-name = "either"
-version = "1.12.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3dca9240753cf90908d7e4aac30f630662b02aebaa1b58a3cadabdb23385b58b"
-
 [[package]]
 name = "enum_delegate"
 version = "0.2.0"
@@ -338,34 +172,6 @@ dependencies = [
  "syn 1.0.109",
 ]
 
-[[package]]
-name = "enum_dispatch"
-version = "0.3.13"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "aa18ce2bc66555b3218614519ac839ddb759a7d6720732f979ef8d13be147ecd"
-dependencies = [
- "once_cell",
- "proc-macro2",
- "quote",
- "syn 2.0.68",
-]
-
-[[package]]
-name = "equivalent"
-version = "1.0.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5443807d6dff69373d433ab9ef5378ad8df50ca6298caf15de6e52e24aaf54d5"
-
-[[package]]
-name = "flate2"
-version = "1.0.30"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5f54427cfd1c7829e2a139fcefea601bf088ebca651d2bf53ebc600eac295dae"
-dependencies = [
- "crc32fast",
- "miniz_oxide",
-]
-
 [[package]]
 name = "form_urlencoded"
 version = "1.2.1"
@@ -431,7 +237,7 @@ checksum = "87750cf4b7a4c0625b1529e4c543c2182106e4dedc60a2a6455e00d212c489ac"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.68",
+ "syn 2.0.72",
 ]
 
 [[package]]
@@ -481,12 +287,6 @@ version = "0.29.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "40ecd4077b5ae9fd2e9e169b102c6c330d0605168eb0e8bf79952b256dbefffd"
 
-[[package]]
-name = "hashbrown"
-version = "0.14.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e5274423e17b7c9fc20b6e7e208532f9b19825d82dfd615708b70edd83df41f1"
-
 [[package]]
 name = "heck"
 version = "0.5.0"
@@ -499,29 +299,6 @@ version = "0.3.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d231dfb89cfffdbc30e7fc41579ed6066ad03abda9e567ccafae602b97ec5024"
 
-[[package]]
-name = "iana-time-zone"
-version = "0.1.60"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e7ffbb5a1b541ea2561f8c41c087286cc091e21e556a4f09a8f6cbf17b69b141"
-dependencies = [
- "android_system_properties",
- "core-foundation-sys",
- "iana-time-zone-haiku",
- "js-sys",
- "wasm-bindgen",
- "windows-core",
-]
-
-[[package]]
-name = "iana-time-zone-haiku"
-version = "0.1.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f31827a206f56af32e590ba56d5d2d085f558508192593743f16b2306495269f"
-dependencies = [
- "cc",
-]
-
 [[package]]
 name = "idna"
 version = "0.5.0"
@@ -533,29 +310,10 @@ dependencies = [
 ]
 
 [[package]]
-name = "indexmap"
-version = "2.2.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "168fb715dda47215e360912c096649d23d58bf392ac62f73919e831745e40f26"
-dependencies = [
- "equivalent",
- "hashbrown",
-]
-
-[[package]]
-name = "is_terminal_polyfill"
-version = "1.70.0"
+name = "indoc"
+version = "2.0.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f8478577c03552c21db0e2724ffb8986a5ce7af88107e6be5d2ee6e158c12800"
-
-[[package]]
-name = "itertools"
-version = "0.10.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b0fd2260e829bddf4cb6ea802289de2f86d6a7a690192fbe91b3f46e0f2c8473"
-dependencies = [
- "either",
-]
+checksum = "b248f5224d1d606005e02c97f5aa4e88eeb230488bcc03bc9ca4d7991399f2b5"
 
 [[package]]
 name = "itoa"
@@ -563,15 +321,6 @@ version = "1.0.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "49f1f14873335454500d59611f1cf4a4b0f786f9ac11f4312a78e4cf2566695b"
 
-[[package]]
-name = "js-sys"
-version = "0.3.69"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "29c15563dc2726973df627357ce0c9ddddbea194836909d655df6a75d2cf296d"
-dependencies = [
- "wasm-bindgen",
-]
-
 [[package]]
 name = "lazy_static"
 version = "1.5.0"
@@ -596,9 +345,9 @@ dependencies = [
 
 [[package]]
 name = "log"
-version = "0.4.21"
+version = "0.4.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "90ed8c1e510134f979dbc4f070f87d4313098b704861a105fe34231c70a3901c"
+checksum = "a7a70ba024b9dc04c27ea2f0c0548feb474ec5c54bba33a7f72f873a39d07b24"
 
 [[package]]
 name = "memchr"
@@ -606,6 +355,15 @@ version = "2.7.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "78ca9ab1a0babb1e7d5695e3530886289c18cf2f87ec19a575a0abdce112e3a3"
 
+[[package]]
+name = "memoffset"
+version = "0.9.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "488016bfae457b036d996092f6cb448677611ce4449e970ceaf42695203f218a"
+dependencies = [
+ "autocfg",
+]
+
 [[package]]
 name = "minimal-lexical"
 version = "0.2.1"
@@ -623,55 +381,45 @@ dependencies = [
 
 [[package]]
 name = "mio"
-version = "0.8.11"
+version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a4a650543ca06a924e8b371db273b2756685faae30f8487da1b56505a8f78b0c"
+checksum = "4569e456d394deccd22ce1c1913e6ea0e54519f577285001215d33557431afe4"
 dependencies = [
+ "hermit-abi",
  "libc",
  "wasi",
- "windows-sys 0.48.0",
+ "windows-sys 0.52.0",
 ]
 
 [[package]]
-name = "nom"
-version = "7.1.3"
+name = "nix"
+version = "0.28.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d273983c5a657a70a3e8f2a01329822f3b8c8172b73826411a55751e404a0a4a"
+checksum = "ab2156c4fce2f8df6c499cc1c763e4394b7482525bf2a9701c9d79d215f519e4"
 dependencies = [
- "memchr",
- "minimal-lexical",
+ "bitflags",
+ "cfg-if",
+ "cfg_aliases",
+ "libc",
 ]
 
 [[package]]
-name = "num"
-version = "0.4.3"
+name = "nom"
+version = "7.1.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "35bd024e8b2ff75562e5f34e7f4905839deb4b22955ef5e73d2fea1b9813cb23"
+checksum = "d273983c5a657a70a3e8f2a01329822f3b8c8172b73826411a55751e404a0a4a"
 dependencies = [
- "num-bigint",
- "num-complex",
- "num-integer",
- "num-iter",
- "num-rational",
- "num-traits",
+ "memchr",
+ "minimal-lexical",
 ]
 
 [[package]]
 name = "num-bigint"
-version = "0.4.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c165a9ab64cf766f73521c0dd2cfdff64f488b8f0b3e621face3462d3db536d7"
-dependencies = [
- "num-integer",
- "num-traits",
-]
-
-[[package]]
-name = "num-complex"
 version = "0.4.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "73f88a1307638156682bada9d7604135552957b7818057dcef22705b4d509495"
+checksum = "a5e44f723f1133c9deac646763579fdb3ac745e418f2a7af9cd0c431da1f20b9"
 dependencies = [
+ "num-integer",
  "num-traits",
 ]
 
@@ -684,28 +432,6 @@ dependencies = [
  "num-traits",
 ]
 
-[[package]]
-name = "num-iter"
-version = "0.1.45"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1429034a0490724d0075ebb2bc9e875d6503c3cf69e235a8941aa757d83ef5bf"
-dependencies = [
- "autocfg",
- "num-integer",
- "num-traits",
-]
-
-[[package]]
-name = "num-rational"
-version = "0.4.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f83d14da390562dca69fc84082e73e548e1ad308d24accdedd2720017cb37824"
-dependencies = [
- "num-bigint",
- "num-integer",
- "num-traits",
-]
-
 [[package]]
 name = "num-traits"
 version = "0.2.19"
@@ -725,22 +451,11 @@ dependencies = [
  "libc",
 ]
 
-[[package]]
-name = "number_range"
-version = "0.3.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "60080faccd4ca50ad0b801b2be686136376b13f691f6eac84817e40973b2e1bb"
-dependencies = [
- "anyhow",
- "itertools",
- "num",
-]
-
 [[package]]
 name = "object"
-version = "0.36.0"
+version = "0.36.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "576dfe1fc8f9df304abb159d767a29d0476f7750fbf8aa7ad07816004a207434"
+checksum = "27b64972346851a39438c60b341ebc01bba47464ae329e55cf343eb93964efd9"
 dependencies = [
  "memchr",
 ]
@@ -751,12 +466,6 @@ version = "1.19.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3fdb12b2476b595f9358c5161aa467c2438859caa136dec86c26fdd2efe17b92"
 
-[[package]]
-name = "openssl-probe"
-version = "0.1.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ff011a302c396a5197692431fc1948019154afc178baf7d8e37367442a4601cf"
-
 [[package]]
 name = "parking_lot"
 version = "0.12.3"
@@ -777,7 +486,7 @@ dependencies = [
  "libc",
  "redox_syscall",
  "smallvec",
- "windows-targets 0.52.5",
+ "windows-targets 0.52.6",
 ]
 
 [[package]]
@@ -798,11 +507,20 @@ version = "0.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184"
 
+[[package]]
+name = "portable-atomic"
+version = "1.7.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "da544ee218f0d287a911e9c99a39a8c9bc8fcad3cb8db5959940044ecfc67265"
+
 [[package]]
 name = "ppv-lite86"
-version = "0.2.17"
+version = "0.2.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
+checksum = "77957b295656769bb8ad2b6a6b09d897d94f05c41b069aede1fcdaa675eaea04"
+dependencies = [
+ "zerocopy",
+]
 
 [[package]]
 name = "proc-macro2"
@@ -814,10 +532,67 @@ dependencies = [
 ]
 
 [[package]]
-name = "queues"
-version = "1.1.0"
+name = "pyo3"
+version = "0.22.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1475abae4f8ad4998590fe3acfe20104f0a5d48fc420c817cd2c09c3f56151f0"
+checksum = "831e8e819a138c36e212f3af3fd9eeffed6bf1510a805af35b0edee5ffa59433"
+dependencies = [
+ "cfg-if",
+ "indoc",
+ "libc",
+ "memoffset",
+ "once_cell",
+ "portable-atomic",
+ "pyo3-build-config",
+ "pyo3-ffi",
+ "pyo3-macros",
+ "unindent",
+]
+
+[[package]]
+name = "pyo3-build-config"
+version = "0.22.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e8730e591b14492a8945cdff32f089250b05f5accecf74aeddf9e8272ce1fa8"
+dependencies = [
+ "once_cell",
+ "target-lexicon",
+]
+
+[[package]]
+name = "pyo3-ffi"
+version = "0.22.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5e97e919d2df92eb88ca80a037969f44e5e70356559654962cbb3316d00300c6"
+dependencies = [
+ "libc",
+ "pyo3-build-config",
+]
+
+[[package]]
+name = "pyo3-macros"
+version = "0.22.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "eb57983022ad41f9e683a599f2fd13c3664d7063a3ac5714cae4b7bee7d3f206"
+dependencies = [
+ "proc-macro2",
+ "pyo3-macros-backend",
+ "quote",
+ "syn 2.0.72",
+]
+
+[[package]]
+name = "pyo3-macros-backend"
+version = "0.22.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ec480c0c51ddec81019531705acac51bcdbeae563557c982aa8263bb96880372"
+dependencies = [
+ "heck",
+ "proc-macro2",
+ "pyo3-build-config",
+ "quote",
+ "syn 2.0.72",
+]
 
 [[package]]
 name = "quote"
@@ -860,10 +635,11 @@ dependencies = [
 
 [[package]]
 name = "redis"
-version = "0.25.4"
+version = "0.26.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e0d7a6955c7511f60f3ba9e86c6d02b3c3f144f8c24b288d1f4e18074ab8bbec"
+checksum = "e902a69d09078829137b4a5d9d082e0490393537badd7c91a3d69d14639e115f"
 dependencies = [
+ "arc-swap",
  "async-trait",
  "bytes",
  "combine",
@@ -872,273 +648,67 @@ dependencies = [
  "futures-util",
  "itoa",
  "log",
+ "num-bigint",
  "percent-encoding",
  "pin-project-lite",
  "rand",
- "rustls",
- "rustls-native-certs",
- "rustls-pemfile",
- "rustls-pki-types",
  "ryu",
  "sha1_smol",
  "socket2",
  "tokio",
- "tokio-rustls",
  "tokio-util",
  "url",
 ]
 
 [[package]]
 name = "redox_syscall"
-version = "0.5.2"
+version = "0.5.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c82cf8cff14456045f55ec4241383baeff27af886adb72ffb2162f99911de0fd"
+checksum = "2a908a6e00f1fdd0dfd9c0eb08ce85126f6d8bbda50017e74bc4a4b7d4a926a4"
 dependencies = [
  "bitflags",
 ]
 
 [[package]]
-name = "regex"
-version = "1.10.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b91213439dad192326a0d7c6ee3955910425f441d7038e0d6933b0aec5c4517f"
-dependencies = [
- "aho-corasick",
- "memchr",
- "regex-automata",
- "regex-syntax",
-]
-
-[[package]]
-name = "regex-automata"
-version = "0.4.7"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "38caf58cc5ef2fed281f89292ef23f6365465ed9a41b7a7754eb4e26496c92df"
-dependencies = [
- "aho-corasick",
- "memchr",
- "regex-syntax",
-]
-
-[[package]]
-name = "regex-syntax"
-version = "0.8.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7a66a03ae7c801facd77a29370b4faec201768915ac14a721ba36f20bc9c209b"
-
-[[package]]
-name = "resp-benchmark"
+name = "resp-benchmark-rust-lib"
 version = "0.1.1"
 dependencies = [
  "awaitgroup",
- "chrono",
- "clap",
  "colored",
  "core_affinity",
+ "ctrlc",
  "enum_delegate",
- "enum_dispatch",
- "futures",
  "nom",
- "num_cpus",
- "number_range",
- "queues",
+ "pyo3",
  "rand",
  "redis",
- "rust_xlsxwriter",
- "serde",
- "serde_json",
  "tokio",
- "toml",
  "zipf",
 ]
 
-[[package]]
-name = "ring"
-version = "0.17.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c17fa4cb658e3583423e915b9f3acc01cceaee1860e33d59ebae66adc3a2dc0d"
-dependencies = [
- "cc",
- "cfg-if",
- "getrandom",
- "libc",
- "spin",
- "untrusted",
- "windows-sys 0.52.0",
-]
-
-[[package]]
-name = "rust_xlsxwriter"
-version = "0.66.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "eea20da37f12cb312effc50d90a8dab9540648972ec6fbd1e3acb9618eb9fb23"
-dependencies = [
- "lazy_static",
- "regex",
- "rust_xlsxwriter_derive",
- "serde",
- "zip",
-]
-
-[[package]]
-name = "rust_xlsxwriter_derive"
-version = "0.2.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "400cc5bbf116143f0ed897d7908b49f604feea981414ef644731e62449d1cc44"
-dependencies = [
- "quote",
- "syn 2.0.68",
-]
-
 [[package]]
 name = "rustc-demangle"
 version = "0.1.24"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "719b953e2095829ee67db738b3bfa9fa368c94900df327b3f07fe6e794d2fe1f"
 
-[[package]]
-name = "rustls"
-version = "0.22.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bf4ef73721ac7bcd79b2b315da7779d8fc09718c6b3d2d1b2d94850eb8c18432"
-dependencies = [
- "log",
- "ring",
- "rustls-pki-types",
- "rustls-webpki",
- "subtle",
- "zeroize",
-]
-
-[[package]]
-name = "rustls-native-certs"
-version = "0.7.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8f1fb85efa936c42c6d5fc28d2629bb51e4b2f4b8a5211e297d599cc5a093792"
-dependencies = [
- "openssl-probe",
- "rustls-pemfile",
- "rustls-pki-types",
- "schannel",
- "security-framework",
-]
-
-[[package]]
-name = "rustls-pemfile"
-version = "2.1.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "29993a25686778eb88d4189742cd713c9bce943bc54251a33509dc63cbacf73d"
-dependencies = [
- "base64",
- "rustls-pki-types",
-]
-
-[[package]]
-name = "rustls-pki-types"
-version = "1.7.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "976295e77ce332211c0d24d92c0e83e50f5c5f046d11082cea19f3df13a3562d"
-
-[[package]]
-name = "rustls-webpki"
-version = "0.102.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ff448f7e92e913c4b7d4c6d8e4540a1724b319b4152b8aef6d4cf8339712b33e"
-dependencies = [
- "ring",
- "rustls-pki-types",
- "untrusted",
-]
-
 [[package]]
 name = "ryu"
 version = "1.0.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f3cb5ba0dc43242ce17de99c180e96db90b235b8a9fdc9543c96d2209116bd9f"
 
-[[package]]
-name = "schannel"
-version = "0.1.23"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fbc91545643bcf3a0bbb6569265615222618bdf33ce4ffbbd13c4bbd4c093534"
-dependencies = [
- "windows-sys 0.52.0",
-]
-
 [[package]]
 name = "scopeguard"
 version = "1.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49"
 
-[[package]]
-name = "security-framework"
-version = "2.11.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c627723fd09706bacdb5cf41499e95098555af3c3c29d014dc3c458ef6be11c0"
-dependencies = [
- "bitflags",
- "core-foundation",
- "core-foundation-sys",
- "libc",
- "security-framework-sys",
-]
-
-[[package]]
-name = "security-framework-sys"
-version = "2.11.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "317936bbbd05227752583946b9e66d7ce3b489f84e11a94a510b4437fef407d7"
-dependencies = [
- "core-foundation-sys",
- "libc",
-]
-
-[[package]]
-name = "serde"
-version = "1.0.203"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7253ab4de971e72fb7be983802300c30b5a7f0c2e56fab8abfc6a214307c0094"
-dependencies = [
- "serde_derive",
-]
-
-[[package]]
-name = "serde_derive"
-version = "1.0.203"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "500cbc0ebeb6f46627f50f3f5811ccf6bf00643be300b4c3eabc0ef55dc5b5ba"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn 2.0.68",
-]
-
-[[package]]
-name = "serde_json"
-version = "1.0.117"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "455182ea6142b14f93f4bc5320a2b31c1f266b66a4a5c858b013302a5d8cbfc3"
-dependencies = [
- "itoa",
- "ryu",
- "serde",
-]
-
-[[package]]
-name = "serde_spanned"
-version = "0.6.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "79e674e01f999af37c49f70a6ede167a8a60b2503e56c5599532a65baa5969a0"
-dependencies = [
- "serde",
-]
-
 [[package]]
 name = "sha1_smol"
-version = "1.0.0"
+version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ae1a47186c03a32177042e55dbc5fd5aee900b8e0069a8d70fba96a9375cd012"
+checksum = "bbfa15b3dddfee50a0fff136974b3e1bde555604ba463834a7eb7deb6417705d"
 
 [[package]]
 name = "signal-hook-registry"
@@ -1174,24 +744,6 @@ dependencies = [
  "windows-sys 0.52.0",
 ]
 
-[[package]]
-name = "spin"
-version = "0.9.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6980e8d7511241f8acf4aebddbb1ff938df5eebe98691418c4468d0b72a96a67"
-
-[[package]]
-name = "strsim"
-version = "0.11.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7da8b5736845d9f2fcb837ea5d9e2628564b3b043a70948a3f0b778838c5fb4f"
-
-[[package]]
-name = "subtle"
-version = "2.6.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0d0208408ba0c3df17ed26eb06992cb1a1268d41b2c0e12e65203fbe3972cee5"
-
 [[package]]
 name = "syn"
 version = "1.0.109"
@@ -1205,20 +757,26 @@ dependencies = [
 
 [[package]]
 name = "syn"
-version = "2.0.68"
+version = "2.0.72"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "901fa70d88b9d6c98022e23b4136f9f3e54e4662c3bc1bd1d84a42a9a0f0c1e9"
+checksum = "dc4b9b9bf2add8093d3f2c0204471e951b2285580335de42f9d2534f3ae7a8af"
 dependencies = [
  "proc-macro2",
  "quote",
  "unicode-ident",
 ]
 
+[[package]]
+name = "target-lexicon"
+version = "0.12.16"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "61c41af27dd6d1e27b1b16b489db798443478cef1f06a660c96db617ba5de3b1"
+
 [[package]]
 name = "tinyvec"
-version = "1.6.0"
+version = "1.8.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "87cc5ceb3875bb20c2890005a4e226a4651264a5c75edb2421b52861a0a0cb50"
+checksum = "445e881f4f6d382d5f27c034e25eb92edd7c784ceab92a0937db7f2e9471b938"
 dependencies = [
  "tinyvec_macros",
 ]
@@ -1231,43 +789,31 @@ checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
 
 [[package]]
 name = "tokio"
-version = "1.38.0"
+version = "1.39.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ba4f4a02a7a80d6f274636f0aa95c7e383b912d41fe721a31f29e29698585a4a"
+checksum = "daa4fb1bc778bd6f04cbfc4bb2d06a7396a8f299dc33ea1900cedaa316f467b1"
 dependencies = [
  "backtrace",
  "bytes",
  "libc",
  "mio",
- "num_cpus",
  "parking_lot",
  "pin-project-lite",
  "signal-hook-registry",
  "socket2",
  "tokio-macros",
- "windows-sys 0.48.0",
+ "windows-sys 0.52.0",
 ]
 
 [[package]]
 name = "tokio-macros"
-version = "2.3.0"
+version = "2.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5f5ae998a069d4b5aba8ee9dad856af7d520c3699e6159b185c2acd48155d39a"
+checksum = "693d596312e88961bc67d7f1f97af8a70227d9f90c31bba5806eec004978d752"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.68",
-]
-
-[[package]]
-name = "tokio-rustls"
-version = "0.25.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "775e0c0f0adb3a2f22a00c4745d728b479985fc15ee7ca6a2608388c5569860f"
-dependencies = [
- "rustls",
- "rustls-pki-types",
- "tokio",
+ "syn 2.0.72",
 ]
 
 [[package]]
@@ -1283,40 +829,6 @@ dependencies = [
  "tokio",
 ]
 
-[[package]]
-name = "toml"
-version = "0.8.14"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6f49eb2ab21d2f26bd6db7bf383edc527a7ebaee412d17af4d40fdccd442f335"
-dependencies = [
- "serde",
- "serde_spanned",
- "toml_datetime",
- "toml_edit",
-]
-
-[[package]]
-name = "toml_datetime"
-version = "0.6.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4badfd56924ae69bcc9039335b2e017639ce3f9b001c393c1b2d1ef846ce2cbf"
-dependencies = [
- "serde",
-]
-
-[[package]]
-name = "toml_edit"
-version = "0.22.14"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f21c7aaf97f1bd9ca9d4f9e73b0a6c74bd5afef56f2bc931943a6e1c37e04e38"
-dependencies = [
- "indexmap",
- "serde",
- "serde_spanned",
- "toml_datetime",
- "winnow",
-]
-
 [[package]]
 name = "unicode-bidi"
 version = "0.3.15"
@@ -1339,10 +851,10 @@ dependencies = [
 ]
 
 [[package]]
-name = "untrusted"
-version = "0.9.0"
+name = "unindent"
+version = "0.2.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8ecb6da28b8a351d773b68d5825ac39017e680750f980f3a1a85cd8dd28a47c1"
+checksum = "c7de7d73e1754487cb58364ee906a499937a0dfabd86bcb980fa99ec8c8fa2ce"
 
 [[package]]
 name = "url"
@@ -1355,72 +867,12 @@ dependencies = [
  "percent-encoding",
 ]
 
-[[package]]
-name = "utf8parse"
-version = "0.2.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821"
-
 [[package]]
 name = "wasi"
 version = "0.11.0+wasi-snapshot-preview1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9c8d87e72b64a3b4db28d11ce29237c246188f4f51057d65a7eab63b7987e423"
 
-[[package]]
-name = "wasm-bindgen"
-version = "0.2.92"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4be2531df63900aeb2bca0daaaddec08491ee64ceecbee5076636a3b026795a8"
-dependencies = [
- "cfg-if",
- "wasm-bindgen-macro",
-]
-
-[[package]]
-name = "wasm-bindgen-backend"
-version = "0.2.92"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "614d787b966d3989fa7bb98a654e369c762374fd3213d212cfc0251257e747da"
-dependencies = [
- "bumpalo",
- "log",
- "once_cell",
- "proc-macro2",
- "quote",
- "syn 2.0.68",
- "wasm-bindgen-shared",
-]
-
-[[package]]
-name = "wasm-bindgen-macro"
-version = "0.2.92"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a1f8823de937b71b9460c0c34e25f3da88250760bec0ebac694b49997550d726"
-dependencies = [
- "quote",
- "wasm-bindgen-macro-support",
-]
-
-[[package]]
-name = "wasm-bindgen-macro-support"
-version = "0.2.92"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e94f17b526d0a461a191c78ea52bbce64071ed5c04c9ffe424dcb38f74171bb7"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn 2.0.68",
- "wasm-bindgen-backend",
- "wasm-bindgen-shared",
-]
-
-[[package]]
-name = "wasm-bindgen-shared"
-version = "0.2.92"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "af190c94f2773fdb3729c55b007a722abb5384da03bc0986df4c289bf5567e96"
-
 [[package]]
 name = "winapi"
 version = "0.3.9"
@@ -1443,15 +895,6 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
 
-[[package]]
-name = "windows-core"
-version = "0.52.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "33ab640c8d7e35bf8ba19b884ba838ceb4fba93a4e8c65a9059d08afcfc683d9"
-dependencies = [
- "windows-targets 0.52.5",
-]
-
 [[package]]
 name = "windows-sys"
 version = "0.48.0"
@@ -1467,7 +910,7 @@ version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "282be5f36a8ce781fad8c8ae18fa3f9beff57ec1b52cb3de0789201425d9a33d"
 dependencies = [
- "windows-targets 0.52.5",
+ "windows-targets 0.52.6",
 ]
 
 [[package]]
@@ -1487,18 +930,18 @@ dependencies = [
 
 [[package]]
 name = "windows-targets"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6f0713a46559409d202e70e28227288446bf7841d3211583a4b53e3f6d96e7eb"
+checksum = "9b724f72796e036ab90c1021d4780d4d3d648aca59e491e6b98e725b84e99973"
 dependencies = [
- "windows_aarch64_gnullvm 0.52.5",
- "windows_aarch64_msvc 0.52.5",
- "windows_i686_gnu 0.52.5",
+ "windows_aarch64_gnullvm 0.52.6",
+ "windows_aarch64_msvc 0.52.6",
+ "windows_i686_gnu 0.52.6",
  "windows_i686_gnullvm",
- "windows_i686_msvc 0.52.5",
- "windows_x86_64_gnu 0.52.5",
- "windows_x86_64_gnullvm 0.52.5",
- "windows_x86_64_msvc 0.52.5",
+ "windows_i686_msvc 0.52.6",
+ "windows_x86_64_gnu 0.52.6",
+ "windows_x86_64_gnullvm 0.52.6",
+ "windows_x86_64_msvc 0.52.6",
 ]
 
 [[package]]
@@ -1509,9 +952,9 @@ checksum = "2b38e32f0abccf9987a4e3079dfb67dcd799fb61361e53e2882c3cbaf0d905d8"
 
 [[package]]
 name = "windows_aarch64_gnullvm"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7088eed71e8b8dda258ecc8bac5fb1153c5cffaf2578fc8ff5d61e23578d3263"
+checksum = "32a4622180e7a0ec044bb555404c800bc9fd9ec262ec147edd5989ccd0c02cd3"
 
 [[package]]
 name = "windows_aarch64_msvc"
@@ -1521,9 +964,9 @@ checksum = "dc35310971f3b2dbbf3f0690a219f40e2d9afcf64f9ab7cc1be722937c26b4bc"
 
 [[package]]
 name = "windows_aarch64_msvc"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9985fd1504e250c615ca5f281c3f7a6da76213ebd5ccc9561496568a2752afb6"
+checksum = "09ec2a7bb152e2252b53fa7803150007879548bc709c039df7627cabbd05d469"
 
 [[package]]
 name = "windows_i686_gnu"
@@ -1533,15 +976,15 @@ checksum = "a75915e7def60c94dcef72200b9a8e58e5091744960da64ec734a6c6e9b3743e"
 
 [[package]]
 name = "windows_i686_gnu"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "88ba073cf16d5372720ec942a8ccbf61626074c6d4dd2e745299726ce8b89670"
+checksum = "8e9b5ad5ab802e97eb8e295ac6720e509ee4c243f69d781394014ebfe8bbfa0b"
 
 [[package]]
 name = "windows_i686_gnullvm"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "87f4261229030a858f36b459e748ae97545d6f1ec60e5e0d6a3d32e0dc232ee9"
+checksum = "0eee52d38c090b3caa76c563b86c3a4bd71ef1a819287c19d586d7334ae8ed66"
 
 [[package]]
 name = "windows_i686_msvc"
@@ -1551,9 +994,9 @@ checksum = "8f55c233f70c4b27f66c523580f78f1004e8b5a8b659e05a4eb49d4166cca406"
 
 [[package]]
 name = "windows_i686_msvc"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db3c2bf3d13d5b658be73463284eaf12830ac9a26a90c717b7f771dfe97487bf"
+checksum = "240948bc05c5e7c6dabba28bf89d89ffce3e303022809e73deaefe4f6ec56c66"
 
 [[package]]
 name = "windows_x86_64_gnu"
@@ -1563,9 +1006,9 @@ checksum = "53d40abd2583d23e4718fddf1ebec84dbff8381c07cae67ff7768bbf19c6718e"
 
 [[package]]
 name = "windows_x86_64_gnu"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4e4246f76bdeff09eb48875a0fd3e2af6aada79d409d33011886d3e1581517d9"
+checksum = "147a5c80aabfbf0c7d901cb5895d1de30ef2907eb21fbbab29ca94c5b08b1a78"
 
 [[package]]
 name = "windows_x86_64_gnullvm"
@@ -1575,9 +1018,9 @@ checksum = "0b7b52767868a23d5bab768e390dc5f5c55825b6d30b86c844ff2dc7414044cc"
 
 [[package]]
 name = "windows_x86_64_gnullvm"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "852298e482cd67c356ddd9570386e2862b5673c85bd5f88df9ab6802b334c596"
+checksum = "24d5b23dc417412679681396f2b49f3de8c1473deb516bd34410872eff51ed0d"
 
 [[package]]
 name = "windows_x86_64_msvc"
@@ -1587,35 +1030,29 @@ checksum = "ed94fce61571a4006852b7389a063ab983c02eb1bb37b47f8272ce92d06d9538"
 
 [[package]]
 name = "windows_x86_64_msvc"
-version = "0.52.5"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bec47e5bfd1bff0eeaf6d8b485cc1074891a197ab4225d504cb7a1ab88b02bf0"
+checksum = "589f6da84c646204747d1270a2a5661ea66ed1cced2631d546fdfb155959f9ec"
 
 [[package]]
-name = "winnow"
-version = "0.6.13"
+name = "zerocopy"
+version = "0.7.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "59b5e5f6c299a3c7890b876a2a587f3115162487e704907d9b6cd29473052ba1"
+checksum = "1b9b4fd18abc82b8136838da5d50bae7bdea537c574d8dc1a34ed098d6c166f0"
 dependencies = [
- "memchr",
+ "byteorder",
+ "zerocopy-derive",
 ]
 
 [[package]]
-name = "zeroize"
-version = "1.8.1"
+name = "zerocopy-derive"
+version = "0.7.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ced3678a2879b30306d323f4542626697a464a97c0a07c9aebf7ebca65cd4dde"
-
-[[package]]
-name = "zip"
-version = "0.6.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "760394e246e4c28189f19d488c058bf16f564016aefac5d32bb1f3b51d5e9261"
+checksum = "fa4f8080344d4671fb4e831a13ad1e68092748387dfc4f55e356242fae12ce3e"
 dependencies = [
- "byteorder",
- "crc32fast",
- "crossbeam-utils",
- "flate2",
+ "proc-macro2",
+ "quote",
+ "syn 2.0.72",
 ]
 
 [[package]]
diff --git a/Cargo.toml b/Cargo.toml
index 0061bd1..a9bbdfa 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,27 +1,29 @@
 [package]
-name = "resp-benchmark"
+name = "resp-benchmark-rust-lib"
 version = "0.1.1"
 edition = "2021"
 description = "A performance benchmarking tool for RESP protocol databases."
 
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+[lib]
+name = "_resp_benchmark_rust_lib"
+crate-type = ["cdylib"]
+
 [dependencies]
+pyo3 = { version = "0.22.2", features = ["extension-module"] }
 tokio = { version = "1", features = ["full"] }
-redis = { version = "0.25", features = ["tokio-comp", "cluster", "tokio-rustls-comp", "tls-rustls-insecure", "cluster-async"] }
-num_cpus = "1.16.0"
-serde = { version = "1.0.193", features = ["derive"] }
-toml = "0.8.8"
+redis = { version = "0.26.1", features = [
+    "tokio-comp",
+    "cluster",
+    "cluster-async",
+#    "tls-native-tls",
+#    "tokio-native-tls-comp",
+] }
 rand = { version = "0.8.5", features = [] }
 zipf = "7.0.1"
-clap = { version = "4.5.4", features = ["derive"] }
 nom = "7.1.3"
-number_range = "0.3.2"
 core_affinity = "0.8.1"
 awaitgroup = "0.7.0"
 colored = "2.1.0"
-rust_xlsxwriter = { version = "0.66.0", features = ["serde"] }
-chrono = "0.4.38"
-queues = "1.1.0"
-futures = "0.3.30"
-enum_dispatch = "0.3.13"
 enum_delegate = "0.2.0"
-serde_json = "1.0.117"
+ctrlc = "3.4.4"
diff --git a/README.md b/README.md
index d133852..9ed45b9 100644
--- a/README.md
+++ b/README.md
@@ -2,128 +2,68 @@
 
 [![License](https://img.shields.io/badge/license-MIT-blue.svg)](https://github.com/your_username/resp-benchmark/blob/main/LICENSE)
 
-`resp-benchmark` is a high-performance benchmarking tool designed for evaluating the performance of various databases
-such as Redis, Valkey, Kvrocks, Tair, and more. It provides a robust and reproducible way to measure the throughput and
-latency of database operations under different workloads.
+用来测试兼容 RESP 协议的数据库，比如 Redis、Valkey、Tair 等，提供命令行工具与 Python 库（方便编写自动化测试脚本）。
 
-![./docs/demo.png](./docs/demo.png)
+## 安装
 
-## Features
-
-- **🔄 Cluster and standalone modes**: Benchmark in both cluster and standalone setups.
-- **⚙️ Adaptive connections**: Auto-adjust connection count based on database performance.
-- **📝 Customizable workloads**: Define your workloads easily with TOML files.
-- **🔧 Placeholders for custom commands**: Easily write performance tests with placeholders.
-- **📊 Detailed output formats**: Export results in JSON and XLSX for analysis.
-- **💡 Efficient resource use**: High performance on smaller machines.
-
-## Getting Started
-
-### Prerequisites
-
-Ensure you have Rust installed. If not, you can install it from [rust-lang.org](https://www.rust-lang.org/).
+```bash
+pip install resp-benchmark
+```
 
-### Installation
+## 使用
 
-Clone the repository and build the project:
+### 命令行工具
 
-```shell
-git clone https://github.com/your_username/resp-benchmark.git
-cd resp-benchmark
-cargo build --release
+```bash
+resp-benchmark --help
 ```
 
-### Running Benchmarks
-
-To run a benchmark, use the following command:
+### Python 库
 
-```shell
-./target/release/resp-benchmark ./workloads/example.toml
-```
+```python
+from resp_benchmark import Benchmark
 
-## Configuration
-
-The benchmark configuration is defined in a TOML file. Below is an example configuration:
-
-```toml
-cluster = false
-address = "127.0.0.1:7001"
-username = ""
-password = ""
-tls = false
-db_type = "redis"  # redis, kvrocks, tair_mem, tair_scm, tair_ssd, garnet
-output = ["xlsx", "json"]
-replica_count = 0
-
-# Leave empty to use all available CPUs. You can specify CPU cores,
-# such as "0,1,6-10" to use cores 0, 1, 6, 7, 8, 9, and 10.
-cpus = ""
-
-[[cases]]
-name = "STRING: SET 64B"
-command = "SET {key uniform 50000000} {value 64}"
-connections = 8
-pipeline = 8
-count = 100_000
-
-[[cases]]
-name = "SET: SISMEMBER"
-dataset = { command = "SADD {key sequence 10} {key sequence 107}", count = 1007_000 }
-command = "SISMEMBER {key uniform 10} {key uniform 1007}"
-connections = 0
-pipeline = 1
-seconds = 10
+bm = Benchmark(host="127.0.0.1", port=6379)
+bm.flushall()
+bm.load_data(command="SET {key sequence 10000000} {value 64}", count=1000_0000, connections=128)
+result = bm.bench("GET {key uniform 10000000}", seconds=3, connections=16)
+print(result.qps, result.avg_latency_ms, result.p99_latency_ms)
 ```
 
-### Field Descriptions
+## 自定义命令
 
-#### Global Configuration
+resp-benchmark 支持自定义要测试的命令，使用如下占位符语法：`SET {key uniform 10000000} {value 64}` 表示执行 SET 命令，key 的分布是 uniform，随机范围是 0-10000000，value 的大小是 64 字节
 
-- **`cluster`**: (Boolean) Specifies if the benchmark should run in cluster mode. Default is `false`.
-- **`address`**: (String) The address of the database server, e.g., `"127.0.0.1:6379"`.
-- **`username`**: (String) The username for database authentication. Leave empty if not required.
-- **`password`**: (String) The password for database authentication. Leave empty if not required.
-- **`tls`**: (Boolean) Enables TLS for secure connections. Default is `false`.
-- **`db_type`**: (String) Specifies the type of database. Options include `redis`, `kvrocks`, `tair_mem`, `tair_scm`, `tair_ssd`, `garnet`.
-- **`output`**: (Array of Strings) Specifies the output formats for the results. Options include `xlsx`, `json`.
+支持的占位符有：
+- **`{key uniform N}`**: 生成范围在 `0` 至 `N-1` 的随机数。比如 `{key uniform 100}` 可能会生成 `key_0000000099`。
+- **`{key sequence N}`**: 同上，但是是顺序产生，用于在加载数据时保证数据覆盖。比如 `{key sequence 100}` 会生成 `key_0000000000`, `key_0000000001`, ...
+- **`{key zipfian N}`**: 同上，但是分布是指数为 1.03 的 Zipfian 分布，用于模拟真实场景下的 key 分布。
+- **`{value N}`**: 生成长度为 `N` 字节的随机字符串。比如 `{value 64}` 可能会生成 `92xsqdNgAyKcqtR4pyXz7j1GQAlRJQJ9TagOmCZ5xR3q3UCXl6B7QysZfgYd4Vmd`。
+- **`{rand N}`**: 生成一个 `0` 到 `N-1` 之间的随机数。比如 `{rand 100}` 可能会生成 `99`。
+- **`{range N W}`**: 生成一对随机数，两个数字的范围是 `0` 到 `N-1`，两个数字的差值是 `W`，用来测试各类 `*range*` 命令。比如 `{range 100 10}` 可能会生成 `89 99`。
 
-#### Resource Configuration
+## 最佳实践
 
-- **`cpus`**: (String) Specifies the CPU cores to use. Leave empty to use all available CPUs. Format: `"0,1,6-10"`.
 
-#### Test Cases
+### 压测 zset 
 
-Each test case is defined in the `[[cases]]` array.
-
-- **`name`**: (String) A descriptive name for the test case.
-- **`command`**: (String) The command to be executed. Supports placeholders for dynamic data generation.
-- **`connections`**: (Integer) The number of concurrent connections. Set to `0` for automatic adjustment.
-- **`pipeline`**: (Integer) The number of commands sent in one batch. Must be `1` in cluster mode.
-- **`count`**: (Integer) The number of operations to perform. Use either `count` or `seconds`, not both.
-- **`seconds`**: (Integer) The duration to run the test case in seconds. Use either `count` or `seconds`, not both.
-- **`dataset`**: (Table) Predefined dataset commands for initializing data before running the test case.
+```shell
+# 1. 加载数据
+resp-benchmark --load -n 1000000 -P 10 "ZADD {key sequence 1000} {rand 1000} {value 8}"
+# 2. 压测
+resp-benchmark "ZRANGEBYSCORE {key uniform 1000} {range 1000 10}"
+```
 
-### Example Case Explanation
+### 压测 lua script
 
-```toml
-[[cases]]
-name = "STRING: SET 64B"
-command = "SET {key uniform 50000000} {value 64}"
-connections = 8
-pipeline = 8
-count = 100_000
+```shell
+redis-cli 'SCRIPT LOAD "return redis.call('\''SET'\'', KEYS[1], ARGV[1])"'
+resp-benchmark "EVALSHA d8f2fad9f8e86a53d2a6ebd960b33c4972cacc37 1 {key uniform 100000} {value 64}"
 ```
 
-- **`name`**: "STRING: SET 64B" – This is a descriptive name for the test.
-- **`command`**: "SET {key uniform 50000000} {value 64}" – Executes a SET command with placeholders for keys and values.
-- **`connections`**: 8 – Uses 8 concurrent connections.
-- **`pipeline`**: 8 – Sends 8 commands in one batch.
-- **`count`**: 100,000 – Executes the command 100,000 times.
-
-### Placeholder Definitions
+## 与 redis-benchmark 的差异
 
-- **`{key uniform N}`**: Generates a random key in the range `0` to `N-1`, formatted as `key_{number}`.
-- **`{key sequence N}`**: Generates sequential keys, cycling through `0` to `N-1`.
-- **`{value N}`**: Generates a random string of length `N` bytes.
-- **`{rand N}`**: Generates a random number between `0` and `N-1`.
-- **`{range N W}`**: Generates random number pairs within `0` to `N-1` with a width `W`, e.g., `0-10`.
+使用 resp-benchmark 与 redis-benchmark 测试 Redis 时，可能会得到不同的结果。常见有以下原因：
+1. redis-benchmark 在测试 set 命令时总是使用相同的 value，这不会导致 DB 的持久化与复制机制。resp-benchmark 则可以使用 `{value 64}` 对每条命令都重新生成数据。
+2. redis-benchmark 在测试 list/set/zset/hash 等命令时，总是使用相同的 primary key，可能会导致多线程 DB 的性能数据失真。resp-benchmark 可以通过 `{key uniform 10000000}` 等占位符生成不同的 key。
+3. redis-benchmark 在集群模式下，发到不同节点的请求都被指定了相同的 slot，可能会导致多线程 DB 的性能数据失真。
\ No newline at end of file
diff --git a/docs/demo.png b/docs/demo.png
deleted file mode 100644
index 1d84b9f..0000000
Binary files a/docs/demo.png and /dev/null differ
diff --git a/pyproject.toml b/pyproject.toml
new file mode 100644
index 0000000..8f61737
--- /dev/null
+++ b/pyproject.toml
@@ -0,0 +1,23 @@
+[build-system]
+requires = ["maturin>=1.7,<2.0"]
+build-backend = "maturin"
+
+[project]
+name = "resp-benchmark"
+requires-python = ">=3.8"
+classifiers = [
+    "Programming Language :: Rust",
+    "Programming Language :: Python :: Implementation :: CPython",
+    "Programming Language :: Python :: Implementation :: PyPy",
+]
+dynamic = ["version"]
+dependencies = ["pydantic"]
+
+[project.scripts]
+# Python entry-point wrapper to be installed in `$venv/bin`
+resp-benchmark = "resp_benchmark.cli:main"  # Python function that uses Rust
+
+[tool.maturin]
+module-name = "resp_benchmark._resp_benchmark_rust_lib"
+features = ["pyo3/extension-module"]
+python-source = "python"
diff --git a/python/resp_benchmark/__init__.py b/python/resp_benchmark/__init__.py
new file mode 100644
index 0000000..0eb9482
--- /dev/null
+++ b/python/resp_benchmark/__init__.py
@@ -0,0 +1 @@
+from .wrapper import Benchmark, Result, ResultPoint
diff --git a/python/resp_benchmark/cli.py b/python/resp_benchmark/cli.py
new file mode 100644
index 0000000..4fe993a
--- /dev/null
+++ b/python/resp_benchmark/cli.py
@@ -0,0 +1,43 @@
+import argparse
+from importlib.metadata import version
+
+from resp_benchmark.wrapper import Benchmark
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(
+        description="RESP Benchmark Tool",
+        add_help=False,
+    )
+
+    parser.add_argument("-h", metavar="host", default="127.0.0.1", help="Server hostname (default 127.0.0.1)")
+    parser.add_argument("-p", metavar="port", type=int, default=6379, help="Server port (default 6379)")
+    parser.add_argument("-u", metavar="username", type=str, default="", help="Used to send ACL style \"AUTH username pass\". Needs -a.")
+    parser.add_argument("-a", metavar="password", type=str, default="", help="Password for Redis Auth")
+    parser.add_argument("-c", metavar="clients", type=int, default=50, help="Number of parallel connections (default 50)")
+    parser.add_argument("--cores", type=str, default=f"", help="Comma-separated list of CPU cores to use.")
+    parser.add_argument("--cluster", action="store_true", help="Enable cluster mode.")
+    parser.add_argument("-n", metavar="requests", type=int, default=100000, help="Total number of requests (default 100000), 0 for unlimited.")
+    parser.add_argument("-s", metavar="seconds", type=int, default=0, help="Total time in seconds (default 0), 0 for unlimited.")
+    parser.add_argument("-P", metavar="pipeline", type=int, default=1, help="Pipeline <numreq> requests. Default 1 (no pipeline).")
+    # parser.add_argument("--tls", action="store_true", help="Use TLS for connection (default false)")
+    parser.add_argument("--load", action="store_true", help="Only load data to Redis, no benchmark.")
+    parser.add_argument('-v', '--version', action='version', version=version('resp_benchmark'))
+    parser.add_argument("--help", action="help", help="Output this help and exit.")
+    parser.add_argument("command", type=str, default="SET {key uniform 100000} {value 64}", nargs="?", help="The Redis command to benchmark (default SET {key uniform 100000} {value 64})")
+
+    args = parser.parse_args()
+    return args
+
+
+def main():
+    args = parse_args()
+    bm = Benchmark(host=args.h, port=args.p, username=args.u, password=args.a, cluster=args.cluster, cores=args.cores, timeout=30)
+    if args.load:
+        bm.load_data(command=args.command, connections=args.c, pipeline=args.P, count=args.n)
+    else:
+        bm.bench(command=args.command, connections=args.c, pipeline=args.P, count=args.n, seconds=args.s)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/python/resp_benchmark/cores.py b/python/resp_benchmark/cores.py
new file mode 100644
index 0000000..dd09b8e
--- /dev/null
+++ b/python/resp_benchmark/cores.py
@@ -0,0 +1,24 @@
+import sys
+from itertools import chain
+from typing import List
+import multiprocessing
+
+
+def parse_range_list(rl):
+    def parse_range(r):
+        if len(r) == 0:
+            return []
+        parts = r.split("-")
+        if len(parts) > 2:
+            raise ValueError("Invalid range: {}".format(r))
+        return range(int(parts[0]), int(parts[-1]) + 1)
+
+    return sorted(set(chain.from_iterable(map(parse_range, rl.split(",")))))
+
+
+def parse_cores_string(cores) -> List[int]:
+    try:
+        return parse_range_list(cores)
+    except ValueError:
+        print(f"Invalid cores range: {cores}.")
+        sys.exit(1)
diff --git a/python/resp_benchmark/wrapper.py b/python/resp_benchmark/wrapper.py
new file mode 100644
index 0000000..8f04418
--- /dev/null
+++ b/python/resp_benchmark/wrapper.py
@@ -0,0 +1,178 @@
+import multiprocessing
+from dataclasses import dataclass
+from typing import List
+
+import pydantic
+import redis
+
+from .cores import parse_cores_string
+
+
+@dataclass
+class ResultPoint:
+    """
+    Represents a single data point in benchmark results.
+
+    Attributes:
+        timestamp_second (int): Unix timestamp in seconds.
+        qps (float): Queries per second at this timestamp.
+        avg_latency_ms (float): Average latency in milliseconds at this timestamp.
+        p99_latency_ms (float): 99th percentile latency in milliseconds at this timestamp.
+    """
+    timestamp_second: int
+    qps: float
+    avg_latency_ms: float
+    p99_latency_ms: float
+
+
+@dataclass
+class Result:
+    """
+    Represents the overall result of a benchmark.
+
+    Attributes:
+        qps (float): Average queries per second.
+        avg_latency_ms (float): Average latency in milliseconds.
+        p99_latency_ms (float): 99th percentile latency in milliseconds.
+        per_second_data (List[ResultPoint]): List of per-second data points.
+    """
+    qps: float
+    avg_latency_ms: float
+    p99_latency_ms: float
+    per_second_data: List[ResultPoint]
+
+
+class Benchmark:
+    """
+    A class to perform and manage benchmark tests on a Redis server.
+
+    Attributes:
+        host (str): The host address of the Redis server.
+        port (int): The port number of the Redis server.
+        username (str): The username for authentication.
+        password (str): The password for authentication.
+        cluster (bool): Whether to connect to a Redis cluster.
+        tls (bool): Whether to use TLS for the connection.
+        timeout (int): Timeout for the connection in seconds.
+        cores (str): Comma-separated list of CPU cores to use.
+    """
+
+    @pydantic.validate_call
+    def __init__(
+            self,
+            host: str = "127.0.0.1",
+            port: int = 6379,
+            username: str = "",
+            password: str = "",
+            cluster: bool = False,
+            # tls: bool = False,
+            timeout: int = 30,
+            cores: str = "",
+    ):
+        self.host = host
+        self.port = port
+        self.username = username
+        self.password = password
+        self.cluster = cluster
+        # self.tls = tls
+        self.timeout = timeout
+        if cores == "":
+            cores = f"0-{multiprocessing.cpu_count() - 1}"
+        self.cores = parse_cores_string(cores)
+
+    def bench(
+            self,
+            command: str,
+            connections: int = 32,
+            pipeline: int = 1,
+            count: int = 100000,
+            seconds: int = 0,
+            quiet: bool = False,
+    ) -> Result:
+        """
+        Runs a benchmark test with the specified parameters.
+
+        Args:
+            command (str): The Redis command to benchmark.
+            connections (int): The number of parallel connections.
+            pipeline (int): The number of commands to pipeline.
+            count (int): The total number of requests to make.
+            seconds (int): The duration of the test in seconds.
+            quiet: (bool): Whether to suppress output.
+        Returns:
+            Result: The results of the benchmark test.
+        """
+        from . import _resp_benchmark_rust_lib
+        ret = _resp_benchmark_rust_lib.benchmark(
+            host=self.host,
+            port=self.port,
+            username=self.username,
+            password=self.password,
+            cluster=self.cluster,
+            tls=False,  # TODO: Implement TLS support
+            timeout=self.timeout,
+            cores=self.cores,
+
+            command=command,
+            connections=connections,
+            pipeline=pipeline,
+            count=count,
+            seconds=seconds,
+            load=False,
+            quiet=quiet,
+        )
+        result = Result(
+            qps=ret.qps,
+            avg_latency_ms=ret.avg_latency_ms,
+            p99_latency_ms=ret.p99_latency_ms,
+            per_second_data=[
+                ResultPoint(
+                    timestamp_second=point.timestamp_second,
+                    qps=point.qps,
+                    avg_latency_ms=point.avg_latency_ms,
+                    p99_latency_ms=point.p99_latency_ms,
+                )
+                for point in ret.per_second_data
+            ],
+        )
+
+        return result
+
+    def load_data(self, command: str, count: int, connections: int = 128, pipeline: int = 10, quiet: bool = False):
+        """
+        Load data into the Redis server using the specified command.
+
+        Args:
+            command (str): The Redis command to use for loading data.
+            count (int): The total number of requests to make.
+            connections (int): The number of parallel connections.
+            pipeline (int): The number of commands to pipeline
+            quiet: (bool): Whether to suppress output.
+        """
+
+        from . import _resp_benchmark_rust_lib
+        _resp_benchmark_rust_lib.benchmark(
+            host=self.host,
+            port=self.port,
+            username=self.username,
+            password=self.password,
+            cluster=self.cluster,
+            tls=self.tls,
+            timeout=self.timeout,
+            cores=self.cores,
+
+            command=command,
+            connections=connections,
+            pipeline=pipeline,
+            count=count,
+            seconds=0,
+            load=True,
+            quiet=quiet,
+        )
+
+    def flushall(self):
+        """
+        Clears all data from all Redis databases.
+        """
+        r = redis.Redis(host=self.host, port=self.port, username=self.username, password=self.password)
+        r.flushall()
diff --git a/src/common/async_flag.rs b/src/async_flag.rs
similarity index 99%
rename from src/common/async_flag.rs
rename to src/async_flag.rs
index 1f73090..139cf1e 100644
--- a/src/common/async_flag.rs
+++ b/src/async_flag.rs
@@ -20,6 +20,7 @@ impl AsyncFlag {
         self.sender.send(true).unwrap();
     }
 }
+
 impl Clone for AsyncFlag {
     fn clone(&self) -> Self {
         Self {
diff --git a/src/limiter.rs b/src/auto_connection.rs
similarity index 52%
rename from src/limiter.rs
rename to src/auto_connection.rs
index dbaf5a7..9b63f5b 100644
--- a/src/limiter.rs
+++ b/src/auto_connection.rs
@@ -4,24 +4,30 @@ use tokio::sync::Notify;
 
 const MAX_CONN: u64 = if cfg!(target_os = "macos") { 64 } else { 1024 }; // 1024 is enough for most cases
 
-pub struct Limiter {
-    pub tcp_conn: u64,
+pub struct ConnLimiter {
+    pub total_conn: u64, // total connection count >= active_conn
     active_conn: AtomicU64,
-    target_conn: AtomicU64,
+    target_conn: AtomicU64, // The target connection count
 
     notify_add: Notify,
 }
 
-impl Limiter {
-    pub fn new(tcp_conn: u64) -> Self {
-        Limiter {
-            tcp_conn,
+impl ConnLimiter {
+    pub fn new(total_conn: u64, target_conn: u64) -> Self {
+        ConnLimiter {
+            total_conn,
             active_conn: AtomicU64::new(0),
-            target_conn: AtomicU64::new(0),
+            target_conn: AtomicU64::new(target_conn),
             notify_add: Notify::new(),
         }
     }
-    pub async fn wait_add(&self) {
+    pub async fn wait_new_conn(&self) {
+        let active_conn = self.active_conn.load(std::sync::atomic::Ordering::SeqCst);
+        let target_conn = self.target_conn.load(std::sync::atomic::Ordering::SeqCst);
+        if active_conn < target_conn {
+            self.active_conn.fetch_add(1, std::sync::atomic::Ordering::SeqCst);
+            return;
+        }
         loop {
             self.notify_add.notified().await;
             let active_conn = self.active_conn.load(std::sync::atomic::Ordering::SeqCst);
@@ -37,17 +43,19 @@ impl Limiter {
         }
     }
     pub fn add_conn(&self) {
-        let old_val = self.target_conn.fetch_add(1, std::sync::atomic::Ordering::SeqCst);
-        if old_val >= self.tcp_conn {
+        let target_conn = self.target_conn.load(std::sync::atomic::Ordering::SeqCst);
+        if target_conn >= self.total_conn {
             return;
         }
+        self.target_conn.fetch_add(1, std::sync::atomic::Ordering::SeqCst);
         loop {
-            self.notify_add.notify_waiters();
+            self.notify_add.notify_one();
             let active_conn = self.active_conn.load(std::sync::atomic::Ordering::SeqCst);
             let target_conn = self.target_conn.load(std::sync::atomic::Ordering::SeqCst);
             if active_conn >= target_conn {
-                break;
+                return;
             }
+            std::thread::sleep(std::time::Duration::from_millis(1));
         }
     }
     pub fn get_active_conn(&self) -> u64 {
@@ -58,48 +66,37 @@ impl Limiter {
     }
 }
 
-pub struct AutoLimiter {
-    pub auto: bool,
+pub struct AutoConnection {
     pub ready: bool,
-    pub limiters: Vec<Arc<Limiter>>,
-    max_conn: u64,
-    target_conn: u64,
+    pub limiters: Vec<Arc<ConnLimiter>>,
 
     last_cnt: u64,
     last_qps: f64,
-    last_last_qps: f64,
     instant: std::time::Instant,
-    step: u64,
     inx: usize,
 }
 
-impl AutoLimiter {
-    pub fn new(mut max_conn: u64, count: u64) -> Self {
+impl AutoConnection {
+    pub fn new(active_conn: u64, thread_count: u64) -> Self {
         let mut limiters = Vec::new();
-        let auto = max_conn == 0;
-        if auto {
-            max_conn = MAX_CONN;
-        }
-        let mut total_connection = max_conn;
-        let mut left_count = count;
-        for _ in 0..count {
+        let auto = active_conn == 0;
+        let mut total_connection = if auto { MAX_CONN } else { active_conn };
+        let mut left_count = thread_count;
+        for _ in 0..thread_count {
             let my_conn = (total_connection + left_count - 1) / left_count;
-            limiters.push(Arc::new(Limiter::new(my_conn)));
+            let target_conn = if auto { 0 } else { my_conn };
+            let conn_limiter = Arc::new(ConnLimiter::new(my_conn, target_conn));
+            limiters.push(conn_limiter);
             total_connection -= my_conn;
             left_count -= 1;
         }
-        AutoLimiter {
-            auto,
-            ready: false,
+        AutoConnection {
+            ready: !auto,
             limiters,
-            max_conn,
-            target_conn: 0,
 
             last_cnt: 0,
             last_qps: 0.0,
-            last_last_qps: 0.0,
             instant: std::time::Instant::now(),
-            step: 1,
             inx: 0,
         }
     }
@@ -111,38 +108,36 @@ impl AutoLimiter {
         self.limiters.iter().map(|limiter| limiter.get_target_conn()).sum()
     }
 
-    pub fn adjust(&mut self, cnt: u64) -> bool {
+    pub fn adjust(&mut self, cnt: u64) {
         if self.ready {
-            return false;
-        }
-        if !self.auto {
-            for _ in 0..self.max_conn {
-                self.limiters[self.inx].add_conn();
-                self.inx = (self.inx + 1) % self.limiters.len();
-                self.target_conn += 1;
-            }
-            self.ready = true;
-            return false;
+            return;
         }
 
-        let target_conn: u64 = self.limiters.iter().map(|limiter| limiter.get_target_conn()).sum();
-        let qps = (cnt - self.last_cnt) as f64 / self.instant.elapsed().as_secs_f64();
-
-        if qps >= self.last_qps * 1.1 && target_conn + self.step * 2 < MAX_CONN {
-            self.step = self.step * 2;
-            self.last_last_qps = self.last_qps;
-            self.last_qps = qps;
+        let elapsed = self.instant.elapsed().as_secs_f64();
+        if elapsed < 0.5 {
+            return;
+        }
+        let qps = (cnt - self.last_cnt) as f64 / elapsed;
+        let need_add_conn;
+        if qps >= self.last_qps * 1.5 || elapsed >= 3f64 {
+            if self.last_qps == 0.0 {
+                need_add_conn = 1; // at least 1 connection
+            } else if qps > self.last_qps * 1.1 {
+                need_add_conn = self.active_conn();
+            } else {
+                self.ready = true;
+                return;
+            }
         } else {
-            self.step = 0;
-            self.ready = true;
-            return false;
+            return;
         }
-        for _ in 0..self.step {
+        for _ in 0..need_add_conn {
             self.limiters[self.inx].add_conn();
             self.inx = (self.inx + 1) % self.limiters.len();
         }
+        self.last_qps = qps;
         self.last_cnt = cnt;
         self.instant = std::time::Instant::now();
-        return true;
+        return;
     }
 }
diff --git a/src/bench.rs b/src/bench.rs
new file mode 100644
index 0000000..8cff62d
--- /dev/null
+++ b/src/bench.rs
@@ -0,0 +1,178 @@
+use std::io::Write;
+use std::sync::Arc;
+use awaitgroup::WaitGroup;
+use colored::Colorize;
+use tokio::{select, task};
+
+use crate::BenchmarkResult;
+use crate::client::ClientConfig;
+use crate::command::Command;
+use crate::auto_connection::{AutoConnection, ConnLimiter};
+use crate::shared_context::SharedContext;
+
+#[derive(Clone)]
+pub struct Case {
+    pub command: Command,
+    pub connections: u64,
+    pub count: u64,
+    pub seconds: u64,
+    pub pipeline: u64,
+}
+
+async fn run_commands_on_single_thread(limiter: Arc<ConnLimiter>, config: ClientConfig, case: Case, context: SharedContext) {
+    let local = task::LocalSet::new();
+    for _ in 0..limiter.total_conn {
+        let limiter = limiter.clone();
+        let config = config.clone();
+        let case = case.clone();
+        let mut context = context.clone();
+        local.spawn_local(async move {
+            let mut client = config.get_client().await;
+            let mut cmd = case.command.clone();
+            let limiter = limiter.clone();
+            select! {
+                _ = limiter.wait_new_conn() =>{}
+                _ = context.wait_stop() => {
+                    return;
+                }
+            }
+            loop {
+                let pipeline_cnt = context.fetch(case.pipeline);
+                if pipeline_cnt == 0 {
+                    context.stop();
+                    break;
+                }
+
+                // prepare pipeline
+                let mut p = Vec::new();
+                for _ in 0..pipeline_cnt {
+                    p.push(cmd.gen_cmd());
+                }
+                let instant = std::time::Instant::now();
+                client.run_commands(p).await;
+                let duration = instant.elapsed().as_micros() as u64;
+                for _ in 0..pipeline_cnt {
+                    context.histogram.record(duration);
+                }
+            }
+        });
+    }
+    local.await;
+}
+
+fn wait_finish(case: &Case, mut auto_connection: AutoConnection, mut context: SharedContext, mut wg: WaitGroup, load: bool, quiet: bool) -> BenchmarkResult {
+    let rt = tokio::runtime::Builder::new_current_thread().enable_all().build().unwrap();
+    let mut result = BenchmarkResult::default();
+
+    rt.block_on(async {
+        let histogram = context.histogram.clone();
+        // calc overall qps
+        let mut overall_time = std::time::Instant::now();
+        let mut overall_cnt_overhead = 0;
+        // for log
+        let mut log_instance = std::time::Instant::now();
+        let mut log_last_cnt = histogram.cnt();
+        let mut interval = tokio::time::interval(std::time::Duration::from_millis(233));
+
+        if auto_connection.ready {
+            context.start_timer();
+            overall_time = std::time::Instant::now();
+            overall_cnt_overhead = 0;
+        }
+
+        loop {
+            select! {
+                _ = interval.tick() => {}
+                _ = wg.wait() => {break;}
+            }
+            {
+                let cnt = histogram.cnt();
+                let qps = (cnt - log_last_cnt) as f64 / log_instance.elapsed().as_secs_f64();
+                let active_conn: u64 = auto_connection.active_conn();
+                let target_conn: u64 = auto_connection.target_conn();
+                if auto_connection.ready {
+                    result.qps = (cnt - overall_cnt_overhead) as f64 / overall_time.elapsed().as_secs_f64();
+                }
+                if !quiet {
+                    if load {
+                        print!("\r\x1B[2KData loading qps: {:.0}, {:.2}%", qps, histogram.cnt() as f64 / case.count as f64 * 100f64);
+                    } else {
+                        print!("\r\x1B[2Kqps: {:.0}(overall {:.0}), active_conn: {}, target_conn: {}, {}", qps, result.qps, active_conn, target_conn, histogram);
+                    }
+                }
+                std::io::stdout().flush().unwrap();
+                log_last_cnt = cnt;
+                log_instance = std::time::Instant::now();
+            }
+            if !auto_connection.ready {
+                let cnt = histogram.cnt();
+                auto_connection.adjust(cnt);
+                if auto_connection.ready {
+                    overall_cnt_overhead = histogram.cnt();
+                    overall_time = std::time::Instant::now();
+                    context.start_timer();
+                }
+            }
+        }
+        let active_conn: u64 = auto_connection.active_conn();
+        if load {
+            print!("\r\x1B[2KData loaded, qps: {:.0}, time elapsed: {:.2}s\n", result.qps, overall_time.elapsed().as_secs_f64());
+        } else {
+            print!("\r\x1B[2Kqps: {:.0}, conn: {}, {}\n", result.qps, active_conn, histogram)
+        };
+        result.avg_latency_ms = histogram.avg() as f64 / 1_000.0;
+        result.p99_latency_ms = histogram.percentile(0.99) as f64 / 1_000.0;
+    });
+    return result;
+}
+
+pub fn do_benchmark(client_config: ClientConfig, cores: Vec<u16>, case: Case, load: bool, quiet: bool) -> BenchmarkResult {
+    if !quiet {
+        println!("{}: {}", "command".bold().blue(), case.command.to_string().green().bold());
+        println!("{}: {}", "connections".bold().blue(), if case.connections == 0 { "auto".to_string() } else { case.connections.to_string() });
+        println!("{}: {}", "count".bold().blue(), case.count);
+        println!("{}: {}", "seconds".bold().blue(), case.seconds);
+        println!("{}: {}", "pipeline".bold().blue(), case.pipeline);
+    }
+
+    // calc connections
+    let auto_connection = AutoConnection::new(case.connections, cores.len() as u64);
+
+    let mut thread_handlers = Vec::new();
+    let wg = WaitGroup::new();
+    let core_ids = core_affinity::get_core_ids().unwrap();
+    let context = SharedContext::new(case.count, case.seconds);
+    for inx in 0..cores.len() {
+        let client_config = client_config.clone();
+        let case = case.clone();
+        let context = context.clone();
+        let wk = wg.worker();
+        let core_id = core_ids[cores[inx] as usize];
+        let limiter = auto_connection.limiters[inx].clone();
+        let thread_handler = std::thread::spawn(move || {
+            core_affinity::set_for_current(core_id); // not work on Apple Silicon
+            let rt = tokio::runtime::Builder::new_current_thread().enable_all().build().unwrap();
+            rt.block_on(async {
+                run_commands_on_single_thread(limiter, client_config, case, context).await;
+                wk.done();
+            });
+        });
+        thread_handlers.push(thread_handler);
+    }
+
+    // log thread
+    let result = wait_finish(&case, auto_connection, context, wg, load, quiet);
+
+    // join all threads
+    for thread_handler in thread_handlers {
+        loop {
+            if thread_handler.is_finished() {
+                thread_handler.join().unwrap();
+                break;
+            }
+            std::thread::sleep(std::time::Duration::from_millis(100));
+        }
+    }
+
+    return result;
+}
\ No newline at end of file
diff --git a/src/client.rs b/src/client.rs
index 795aafc..854e781 100644
--- a/src/client.rs
+++ b/src/client.rs
@@ -2,52 +2,17 @@ use enum_delegate;
 use redis::aio::ConnectionLike;
 use redis::aio::MultiplexedConnection;
 use redis::cluster_async::ClusterConnection;
-use redis::cluster_routing::{RoutingInfo, SingleNodeRoutingInfo};
-use redis::{from_redis_value, Cmd, InfoDict, RedisFuture, Value};
-use serde::Deserialize;
+use redis::{Cmd, RedisFuture, Value};
 use std::fmt::{Display, Formatter};
-#[derive(Debug, Clone, Deserialize, PartialEq, Eq, Default)]
-#[serde(rename_all = "snake_case")]
-pub enum DBType {
-    // https://github.com/redis/redis
-    #[default]
-    Redis,
 
-    // https://github.com/valkey-io/valkey
-    Valkey,
-
-    // https://www.alibabacloud.com/product/tair
-    #[serde(rename = "tair_mem")]
-    TairMem,
-    #[serde(rename = "tair_scm")]
-    TairScm,
-    #[serde(rename = "tair_ssd")]
-    TairSsd,
-
-    Garnet,
-
-    // Others
-    #[serde(untagged)]
-    Unknown(String),
-}
-
-#[derive(Deserialize, Clone, Debug)]
-#[serde(deny_unknown_fields)]
+#[derive(Clone)]
 pub struct ClientConfig {
-    #[serde(default)]
     pub cluster: bool,
-    #[serde(default)]
     pub address: String,
-    #[serde(default)]
     pub username: String,
-    #[serde(default)]
     pub password: String,
-    #[serde(default)]
     pub tls: bool,
-    #[serde(default)]
-    pub db_type: DBType,
-    #[serde(default)]
-    replica_count: u64,
+    pub timeout: u64,
 }
 
 impl ClientConfig {
@@ -57,9 +22,10 @@ impl ClientConfig {
         } else {
             format!("redis://{}:{}@{}", &self.username, &self.password, &self.address)
         };
-        return if self.cluster {
+
+        if self.cluster {
             let nodes = vec![conn_str];
-            let client = redis::cluster::ClusterClient::builder(nodes).connection_timeout(std::time::Duration::from_secs(5)).build().unwrap();
+            let client = redis::cluster::ClusterClient::builder(nodes).connection_timeout(std::time::Duration::from_secs(self.timeout)).build().unwrap();
             let conn = match client.get_async_connection().await {
                 Ok(conn) => conn,
                 Err(e) => {
@@ -67,7 +33,7 @@ impl ClientConfig {
                     std::process::exit(1);
                 }
             };
-            Client::new(conn.into(), self.db_type.clone(), self.replica_count)
+            Client::new(conn.into())
         } else {
             let client = redis::Client::open(conn_str).unwrap();
             let conn = match client.get_multiplexed_async_connection().await {
@@ -77,8 +43,8 @@ impl ClientConfig {
                     std::process::exit(1);
                 }
             };
-            Client::new(conn.into(), self.db_type.clone(), self.replica_count)
-        };
+            Client::new(conn.into())
+        }
     }
 }
 
@@ -89,16 +55,16 @@ impl Display for ClientConfig {
 }
 
 #[enum_delegate::implement(ConnectionLike,
-    trait ConnectionLike {
-        fn req_packed_command<'a>(&'a mut self, cmd: &'a Cmd) -> RedisFuture<'a, Value>;
-        fn req_packed_commands<'a>(
-        &'a mut self,
-        cmd: &'a redis::Pipeline,
-        offset: usize,
-        count: usize,
-        ) -> RedisFuture<'a, Vec<Value>>;
-        fn get_db(&self) -> i64;
-    }
+trait ConnectionLike {
+fn req_packed_command < 'a > (& 'a mut self, cmd: & 'a Cmd) -> RedisFuture < 'a, Value >;
+fn req_packed_commands < 'a > (
+& 'a mut self,
+cmd: & 'a redis::Pipeline,
+offset: usize,
+count: usize,
+) -> RedisFuture < 'a, Vec < Value >>;
+fn get_db(& self) -> i64;
+}
 )]
 enum ClientConnection {
     Standalone(MultiplexedConnection),
@@ -107,56 +73,13 @@ enum ClientConnection {
 
 pub struct Client {
     conn: ClientConnection,
-    db_type: DBType,
-    replica_count: u64,
 }
 
 impl Client {
-    fn new(conn: ClientConnection, db_type: DBType, replica_count: u64) -> Client {
-        Client { conn, db_type, replica_count }
-    }
-
-    pub async fn flushall(&mut self) {
-        // flushall
-        let cmd = {
-            let mut cmd = Cmd::new();
-            match &self.db_type {
-                DBType::Garnet => cmd.arg("FLUSHDB"),
-                DBType::TairScm => cmd.arg("FLUSHALL"),
-                _ => cmd.arg("FLUSHALL"),
-            };
-            cmd
-        };
-        match cmd.query_async(&mut self.conn).await {
-            Ok(()) => {}
-            Err(e) => {
-                eprintln!("Failed to execute `flushall` command: {:?}", e);
-                std::process::exit(1);
-            }
-        }
-
-        // wait
-        let mut cmd = Cmd::new();
-        cmd.arg("WAIT").arg(self.replica_count).arg(0);
-        match cmd.query_async(&mut self.conn).await {
-            Ok(()) => {}
-            Err(e) => {
-                eprintln!("Failed to execute `wait` command: {:?}", e);
-                std::process::exit(1);
-            }
-        }
+    fn new(conn: ClientConnection) -> Client {
+        Client { conn }
     }
 
-    async fn info(&mut self, key: &str) -> InfoDict {
-        match self.conn {
-            ClientConnection::Standalone(ref mut conn) => redis::cmd("INFO").arg(key).query_async(conn).await.unwrap(),
-            ClientConnection::Cluster(ref mut conn) => {
-                let random = RoutingInfo::SingleNode(SingleNodeRoutingInfo::Random);
-                let value = conn.route_command(&mut redis::cmd("INFO").arg(key), random).await.unwrap();
-                from_redis_value(&value).unwrap()
-            }
-        }
-    }
 
     pub async fn run_commands(&mut self, cmds: Vec<redis::Cmd>) {
         let mut pipeline = redis::pipe();
@@ -171,28 +94,4 @@ impl Client {
             }
         }
     }
-
-    /// return the `used_memory` in bytes
-    pub async fn info_memory(&mut self) -> u64 {
-        let (info_pkey, info_skey) = match &self.db_type {
-            DBType::Redis => ("MEMORY", "used_memory"),
-            DBType::Valkey => ("MEMORY", "used_memory"),
-            DBType::Garnet => ("MEMORY", "proc_physical_memory_size"),
-            DBType::TairMem => ("MEMORY", "used_memory"),
-            DBType::TairScm => ("PERSISTENCE", "used_pmem"),
-            DBType::TairSsd => ("PERSISTENCE", "data_used_disk_size"),
-            DBType::Unknown(db) => {
-                eprintln!("Unknown db type: {}", db);
-                std::process::exit(1);
-            }
-        };
-        let info: InfoDict = self.info(info_pkey).await;
-        return match info.get(info_skey) {
-            Some::<u64>(used_memory) => used_memory,
-            None => {
-                eprintln!("Failed to get `used_memory` from `info memory` command: {:?}", info);
-                std::process::exit(1);
-            }
-        };
-    }
 }
diff --git a/src/command/mod.rs b/src/command/mod.rs
index 1eb62a4..392386c 100644
--- a/src/command/mod.rs
+++ b/src/command/mod.rs
@@ -10,6 +10,7 @@ mod placeholder;
 pub struct Command {
     str: String,
     argv: Vec<PlaceholderEnum>,
+    #[allow(dead_code)]
     lock: Arc<Mutex<()>>,
 }
 
@@ -39,6 +40,7 @@ impl Command {
         }
         cmd
     }
+    #[allow(dead_code)]
     pub fn gen_cmd_with_lock(&mut self) -> redis::Cmd {
         let mut cmd = redis::Cmd::new();
         let _lock = self.lock.lock().unwrap();
diff --git a/src/common/mod.rs b/src/common/mod.rs
deleted file mode 100644
index 2d97c53..0000000
--- a/src/common/mod.rs
+++ /dev/null
@@ -1 +0,0 @@
-pub(crate) mod async_flag;
diff --git a/src/config.rs b/src/config.rs
deleted file mode 100644
index 0f4b6d4..0000000
--- a/src/config.rs
+++ /dev/null
@@ -1,168 +0,0 @@
-use crate::client::ClientConfig;
-use crate::command::Command;
-use clap::{command, Parser};
-use number_range::NumberRangeOptions;
-use serde::{Deserialize, Deserializer};
-use std::process::exit;
-
-#[derive(Deserialize, Clone, Debug)]
-#[serde(deny_unknown_fields)]
-pub struct Dataset {
-    pub command: Command,
-    pub count: u64,
-}
-
-impl<'de> Deserialize<'de> for Command {
-    fn deserialize<D>(deserializer: D) -> Result<Command, D::Error>
-    where
-        D: Deserializer<'de>,
-    {
-        let s = String::deserialize(deserializer)?;
-        Ok(Command::new(s.as_str()))
-    }
-}
-const fn _default_1() -> u64 {
-    1
-}
-
-#[derive(Deserialize, Clone, Debug)]
-#[serde(deny_unknown_fields)]
-pub struct Case {
-    pub name: Option<String>,
-    pub dataset: Option<Dataset>,
-    pub command: Command,
-    pub connections: u64,
-    #[serde(default)]
-    pub count: u64,
-    #[serde(default)]
-    pub seconds: u64,
-    #[serde(default = "_default_1")]
-    pub pipeline: u64,
-}
-
-impl Default for Case {
-    fn default() -> Self {
-        Case {
-            name: None,
-            dataset: None,
-            command: Command::new("PING"),
-            connections: 0,
-            count: 0,
-            seconds: 0,
-            pipeline: 1,
-        }
-    }
-}
-
-#[derive(Debug, Clone, Deserialize, PartialEq, Eq, Default)]
-#[serde(rename_all = "snake_case")]
-pub enum OutputFormat {
-    #[default]
-    #[serde(rename = "xlsx")]
-    XLSX,
-
-    #[serde(rename = "json")]
-    JSON,
-}
-
-#[derive(Deserialize, Clone, Debug)]
-#[serde(deny_unknown_fields)]
-pub struct Config {
-    #[serde(flatten)]
-    pub client_config: ClientConfig,
-    #[serde(default)]
-    pub cpus: CPUS,
-    #[serde(default)]
-    #[serde(rename = "output")]
-    pub output_formats: Vec<OutputFormat>,
-    #[serde(default)]
-    pub cases: Vec<Case>,
-}
-
-#[derive(Debug, Clone)]
-pub struct CPUS(Vec<u64>);
-
-impl Default for CPUS {
-    fn default() -> Self {
-        let vec: Vec<u64> = (0..num_cpus::get() as u64).collect();
-        CPUS(vec)
-    }
-}
-
-impl CPUS {
-    pub fn iter(&self) -> std::slice::Iter<u64> {
-        self.0.iter()
-    }
-    pub fn get(&self, idx: usize) -> u64 {
-        self.0[idx]
-    }
-    pub fn len(&self) -> usize {
-        self.0.len()
-    }
-}
-
-impl<'de> Deserialize<'de> for CPUS {
-    fn deserialize<D>(deserializer: D) -> Result<CPUS, D::Error>
-    where
-        D: Deserializer<'de>,
-    {
-        let s = String::deserialize(deserializer)?;
-        if s.is_empty() {
-            let vec: Vec<u64> = (0..num_cpus::get() as u64).collect();
-            Ok(CPUS(vec))
-        } else {
-            let vec: Vec<u64> = NumberRangeOptions::new().with_list_sep(',').with_range_sep('-').parse(s.as_str()).unwrap().collect();
-            Ok(CPUS(vec))
-        }
-    }
-}
-
-const AFTER_HELP: &str = r#""#;
-
-#[derive(Parser, Debug, Clone)]
-#[command(version, about, long_about = None, after_help = AFTER_HELP)]
-struct Cli {
-    /// Path to the workload file, e.g., `./workload/redis.toml`
-    workload: String,
-}
-
-impl Config {
-    pub fn parse() -> Config {
-        let cli = Cli::parse();
-        return Config::from_file(&cli.workload);
-    }
-
-    pub fn from_file(path: &str) -> Config {
-        let contents = match std::fs::read_to_string(path) {
-            Ok(c) => c,
-            Err(e) => {
-                eprintln!("Could not read file `{}`: {}", path, e);
-                exit(1);
-            }
-        };
-
-        let conf: Config = match toml::from_str(&contents) {
-            Ok(c) => c,
-            Err(e) => {
-                eprintln!("Could not parse file `{}`: {}", path, e);
-                exit(1);
-            }
-        };
-        return conf;
-    }
-
-    pub fn client_config(&self) -> ClientConfig {
-        self.client_config.clone()
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use number_range::NumberRangeOptions;
-
-    #[test]
-    fn test() {
-        let v: Vec<usize> = NumberRangeOptions::new().with_list_sep(',').with_range_sep('-').parse("1,3-10,14").unwrap().collect();
-        println!("{:?}", v);
-    }
-}
diff --git a/src/counter.rs b/src/counter.rs
deleted file mode 100644
index 10fae63..0000000
--- a/src/counter.rs
+++ /dev/null
@@ -1,31 +0,0 @@
-use std::cmp::min;
-use std::sync::atomic::AtomicU64;
-use tokio::sync::Notify;
-
-pub(crate) struct Counter {
-    current_count: AtomicU64,
-    max_count: u64,
-    stop_notify: Notify,
-}
-impl Counter {
-    pub fn new(max_count: u64) -> Self {
-        Counter {
-            current_count: AtomicU64::new(0),
-            max_count,
-            stop_notify: Notify::new(),
-        }
-    }
-    pub fn stop(&self) {
-        self.stop_notify.notify_waiters();
-    }
-    pub async fn wait_stop(&self) {
-        self.stop_notify.notified().await;
-    }
-    pub fn fetch(&self, count: u64) -> u64 {
-        let prev_count = self.current_count.fetch_add(count, std::sync::atomic::Ordering::Relaxed);
-        if prev_count >= self.max_count {
-            return 0;
-        }
-        return min(self.max_count - prev_count, count);
-    }
-}
diff --git a/src/histogram.rs b/src/histogram.rs
index 311c1b0..815556b 100644
--- a/src/histogram.rs
+++ b/src/histogram.rs
@@ -38,6 +38,7 @@ impl Histogram {
         self.buckets[index as usize].fetch_add(1, Ordering::Relaxed);
     }
 
+    #[allow(dead_code)]
     pub fn clear(&self) {
         self.cnt.store(0, Ordering::Relaxed);
         for i in 0..self.buckets.len() {
@@ -45,6 +46,7 @@ impl Histogram {
         }
     }
 
+    #[allow(dead_code)]
     pub fn un_record(&self, latency_us: u64) {
         let index = match latency_us {
             0..=999 => latency_us / 10,                            // <1ms precision 10us
@@ -122,9 +124,8 @@ impl Display for Histogram {
         }
         let avg = self.avg();
         let p99 = self.percentile(0.99);
-        let p999 = self.percentile(0.999);
 
-        write!(f, "cnt: {}, avg: {}, p99: {}, p999: {}", cnt, Histogram::humanize_us(avg), Histogram::humanize_us(p99), Histogram::humanize_us(p999))
+        write!(f, "cnt: {}, avg: {}, p99: {}", cnt, Histogram::humanize_us(avg), Histogram::humanize_us(p99))
     }
 }
 
diff --git a/src/lib.rs b/src/lib.rs
new file mode 100644
index 0000000..839fc60
--- /dev/null
+++ b/src/lib.rs
@@ -0,0 +1,85 @@
+mod bench;
+mod client;
+mod command;
+mod auto_connection;
+mod shared_context;
+mod histogram;
+mod async_flag;
+
+use ctrlc;
+use pyo3::prelude::*;
+use pyo3::wrap_pyfunction;
+use crate::command::Command;
+
+/// A Python module implemented in Rust.
+#[pymodule]
+fn _resp_benchmark_rust_lib(m: &Bound<'_, PyModule>) -> PyResult<()> {
+    m.add_function(wrap_pyfunction!(benchmark, m)?)?;
+    Ok(())
+}
+
+#[pyclass]
+#[derive(Clone)]
+struct ResultPoint {
+    #[pyo3(get, set)] pub timestamp_second: i64,
+    #[pyo3(get, set)] pub qps: f64,
+    #[pyo3(get, set)] pub avg_latency_ms: f64,
+    #[pyo3(get, set)] pub p99_latency_ms: f64,
+}
+
+#[pyclass]
+#[derive(Default)]
+struct BenchmarkResult {
+    #[pyo3(get, set)] pub qps: f64,
+    #[pyo3(get, set)] pub avg_latency_ms: f64,
+    #[pyo3(get, set)] pub p99_latency_ms: f64,
+    #[pyo3(get, set)] pub per_second_data: Vec<ResultPoint>,
+}
+
+#[pyfunction]
+fn benchmark(
+    host: String,
+    port: u16,
+    username: String,
+    password: String,
+    cluster: bool,
+    tls: bool,
+    timeout: u64,
+    cores: Vec<u16>,
+    command: String,
+    connections: u64,
+    pipeline: u64,
+    count: u64,
+    seconds: u64,
+    load: bool,
+    quiet: bool,
+) -> PyResult<BenchmarkResult> {
+    assert!(cores.len() > 0);
+    if load {
+        assert_ne!(connections, 0);
+        assert_ne!(count, 0);
+    }
+    assert!(count != 0 || seconds != 0);
+
+    let _ = ctrlc::set_handler(move || {
+        std::process::exit(0);
+    });
+
+    let client_config = client::ClientConfig {
+        cluster,
+        address: format!("{}:{}", host, port),
+        username,
+        password,
+        tls,
+        timeout,
+    };
+    let case = bench::Case {
+        command: Command::new(command.as_str()),
+        connections,
+        pipeline,
+        count,
+        seconds,
+    };
+    let result = bench::do_benchmark(client_config, cores, case, load, quiet);
+    Ok(result)
+}
\ No newline at end of file
diff --git a/src/main.rs b/src/main.rs
deleted file mode 100644
index 960a2dd..0000000
--- a/src/main.rs
+++ /dev/null
@@ -1,264 +0,0 @@
-mod client;
-mod command;
-mod common;
-mod config;
-mod histogram;
-mod limiter;
-mod output;
-mod shared_context;
-
-use crate::client::ClientConfig;
-use crate::config::{Case, Dataset};
-use crate::limiter::{AutoLimiter, Limiter};
-use crate::shared_context::SharedContext;
-use awaitgroup::WaitGroup;
-use colored::Colorize;
-use core_affinity;
-pub use histogram::Histogram;
-use std::cmp::min;
-use std::io::Write;
-use std::sync::atomic::AtomicU64;
-use std::sync::Arc;
-use tokio::task::JoinSet;
-use tokio::{select, task};
-
-async fn load_data(config: ClientConfig, dataset: Dataset, total_count: Arc<AtomicU64>, conn_per_thread: usize) {
-    let mut tasks = JoinSet::new();
-    let pipeline_count = if config.cluster { 1 } else { 8 };
-    for inx in 0..conn_per_thread {
-        let config = config.clone();
-        let mut cmd = dataset.command.clone();
-        let total_count = total_count.clone();
-        let dataset = dataset.clone();
-        tasks.spawn(async move {
-            let mut client = config.get_client().await;
-            loop {
-                let prev_count = total_count.fetch_add(pipeline_count, std::sync::atomic::Ordering::SeqCst);
-                if prev_count >= dataset.count {
-                    break;
-                }
-                if inx == 0 && prev_count % 1007 == 0 {
-                    print!("\r\x1B[2K{}: {} {}/{}", "dataset loading".blue().bold(), dataset.command.to_string().green().bold(), prev_count, dataset.count);
-                    std::io::stdout().flush().unwrap();
-                }
-                let pipeline_cnt = min(dataset.count - prev_count, pipeline_count);
-                let mut pipeline = Vec::new();
-                for _ in 0..pipeline_cnt {
-                    pipeline.push(cmd.gen_cmd_with_lock());
-                }
-                client.run_commands(pipeline).await;
-            }
-        });
-    }
-    for _ in 0..conn_per_thread {
-        tasks.join_next().await;
-    }
-}
-
-async fn run_commands_on_single_thread(limiter: Arc<Limiter>, config: ClientConfig, case: Case, context: SharedContext) {
-    let local = task::LocalSet::new();
-    for _ in 0..limiter.tcp_conn {
-        let limiter = limiter.clone();
-        let config = config.clone();
-        let case = case.clone();
-        let mut context = context.clone();
-        local.spawn_local(async move {
-            let mut client = config.get_client().await;
-            let mut cmd = case.command.clone();
-            let limiter = limiter.clone();
-            select! {
-                _ = limiter.wait_add() =>{}
-                _ = context.wait_stop() => {
-                    return;
-                }
-            }
-            loop {
-                let pipeline_cnt = context.fetch(case.pipeline);
-                if pipeline_cnt == 0 {
-                    context.stop();
-                    break;
-                }
-
-                // prepare pipeline
-                let mut p = Vec::new();
-                for _ in 0..pipeline_cnt {
-                    p.push(cmd.gen_cmd());
-                }
-                let instant = std::time::Instant::now();
-                client.run_commands(p).await;
-                let duration = instant.elapsed().as_micros() as u64;
-                for _ in 0..pipeline_cnt {
-                    context.histogram.record(duration);
-                }
-            }
-        });
-    }
-    local.await;
-}
-
-fn wait_finish(case: &Case, client_config: &ClientConfig, mut auto_limiter: AutoLimiter, mut context: SharedContext, mut wg: WaitGroup) -> output::Result {
-    let rt = tokio::runtime::Builder::new_current_thread().enable_all().build().unwrap();
-    let mut result = output::Result::default();
-    result.name = case.name.clone().unwrap_or(case.command.to_string());
-    rt.block_on(async {
-        let histogram = context.histogram.clone();
-        let mut client = client_config.get_client().await;
-        // calc overall qps
-        let mut overall_time = std::time::Instant::now();
-        let mut overall_cnt_overhead = 0;
-        // for log
-        let mut log_instance = std::time::Instant::now();
-        let mut log_last_cnt = histogram.cnt();
-        let mut log_interval = tokio::time::interval(std::time::Duration::from_millis(100));
-        // for auto limiter
-        let mut auto_interval = tokio::time::interval(std::time::Duration::from_secs(10));
-
-        let mut wake_log = false;
-        let mut wake_auto = false;
-        loop {
-            select! {
-                _ = log_interval.tick() => {wake_log=true;}
-                _ = auto_interval.tick() => {wake_auto=true;}
-                _ = wg.wait() => {break;}
-            }
-            if wake_log {
-                wake_log = false;
-                result.memory = client.info_memory().await as f64 / 1024.0 / 1024.0 / 1024.0;
-                let cnt = histogram.cnt();
-                let qps = (cnt - log_last_cnt) as f64 / log_instance.elapsed().as_secs_f64();
-                let active_conn: u64 = auto_limiter.active_conn();
-                let target_conn: u64 = auto_limiter.target_conn();
-                if auto_limiter.ready {
-                    result.qps = (cnt - overall_cnt_overhead) as f64 / overall_time.elapsed().as_secs_f64();
-                }
-                print!("\r\x1B[2Kqps: {:.0}(overall {:.0}), active_conn: {}, target_conn: {}, {}, mem: {:.3}GiB", qps, result.qps, active_conn, target_conn, histogram, result.memory);
-                std::io::stdout().flush().unwrap();
-                log_last_cnt = cnt;
-                log_instance = std::time::Instant::now();
-            }
-            if wake_auto && !auto_limiter.ready {
-                wake_auto = false;
-
-                let cnt = histogram.cnt();
-                if !auto_limiter.adjust(cnt) {
-                    overall_cnt_overhead = histogram.cnt();
-                    overall_time = std::time::Instant::now();
-                    context.start_timer();
-                }
-            }
-        }
-        let active_conn: u64 = auto_limiter.active_conn();
-        print!("\r\x1B[2Kqps: {:.0}, conn: {}, {}\n", result.qps, active_conn, histogram);
-        result.avg = histogram.avg() as f64 / 1_000.0;
-        result.min = histogram.percentile(0.0) as f64 / 1_000.0;
-        result.p50 = histogram.percentile(0.5) as f64 / 1_000.0;
-        result.p90 = histogram.percentile(0.9) as f64 / 1_000.0;
-        result.p99 = histogram.percentile(0.99) as f64 / 1_000.0;
-        result.max = histogram.percentile(1.0) as f64 / 1_000.0;
-        result.connections = active_conn;
-        result.pipeline = case.pipeline;
-        result.count = histogram.cnt();
-        result.duration = overall_time.elapsed().as_secs_f64();
-    });
-    return result;
-}
-
-fn main() {
-    let conf = config::Config::parse();
-    println!("{}: {:?}", "CPU Core".bold(), conf.cpus);
-    println!("{}: {}", "Address".bold(), conf.client_config.address);
-    println!("{}: {}", "Cluster".bold(), conf.client_config.cluster);
-    println!("{}: {}", "TLS".bold(), conf.client_config.tls);
-    println!("{}: {:?}", "output".bold(), conf.output_formats);
-
-    // check cpu core
-    let core_ids = core_affinity::get_core_ids().unwrap();
-    for cpu_id in conf.cpus.iter() {
-        if *cpu_id as usize >= core_ids.len() {
-            // print red
-            println!("{}: {:?}", "Invalid CPU core ids".bold().red(), conf.cpus);
-            std::process::exit(1);
-        }
-    }
-
-    let mut results = output::Results::new();
-    for case in conf.cases.iter() {
-        println!();
-        if let Some(name) = case.name.clone() {
-            println!("{}: {}", "name".bold().blue(), name.bold());
-        }
-
-        // flushall
-        let rt = tokio::runtime::Builder::new_current_thread().enable_all().build().unwrap();
-        rt.block_on(async {
-            let mut client = conf.client_config.get_client().await;
-            client.flushall().await;
-        });
-
-        if let Some(dataset) = &case.dataset {
-            let mut thread_handlers = Vec::new();
-            let core_ids = core_affinity::get_core_ids().unwrap();
-            let count = Arc::new(AtomicU64::new(0));
-            let conn_per_thread = if cfg!(target_os = "macos") { 4 } else { min(64, 512 / conf.cpus.len()) };
-            for cpu_id in conf.cpus.iter() {
-                let core_id = core_ids[*cpu_id as usize];
-                let redis_config = conf.client_config();
-                let dataset = dataset.clone();
-                let count = count.clone();
-                let thread_handler = std::thread::spawn(move || {
-                    core_affinity::set_for_current(core_id); // not work on Apple Silicon
-                    let rt = tokio::runtime::Builder::new_current_thread().enable_all().build().unwrap();
-                    rt.block_on(async {
-                        load_data(redis_config, dataset, count, conn_per_thread).await;
-                    });
-                });
-                thread_handlers.push(thread_handler);
-            }
-            for thread_handler in thread_handlers {
-                thread_handler.join().unwrap();
-            }
-            println!("\r\x1B[2K{}: {} x {}", "dataset".blue().bold(), dataset.command.to_string().bold(), dataset.count);
-        }
-
-        println!("{}: {}", "command".bold().blue(), case.command.to_string().green().bold());
-        println!("{}: {}", "connections".bold().blue(), if case.connections == 0 { "auto".to_string() } else { case.connections.to_string() });
-        println!("{}: {}", "count".bold().blue(), case.count);
-        println!("{}: {}", "pipeline".bold().blue(), case.pipeline);
-
-        // calc connections
-        let auto_limiter = AutoLimiter::new(case.connections, conf.cpus.len() as u64);
-
-        let mut thread_handlers = Vec::new();
-        let wg = WaitGroup::new();
-        let core_ids = core_affinity::get_core_ids().unwrap();
-        let context = SharedContext::new(case.count, case.seconds);
-        for inx in 0..conf.cpus.len() {
-            let client_config = conf.client_config().clone();
-            let case = case.clone();
-            let context = context.clone();
-            let wk = wg.worker();
-            let core_id = core_ids[conf.cpus.get(inx) as usize];
-            let limiter = auto_limiter.limiters[inx].clone();
-            let thread_handler = std::thread::spawn(move || {
-                core_affinity::set_for_current(core_id); // not work on Apple Silicon
-                let rt = tokio::runtime::Builder::new_current_thread().enable_all().build().unwrap();
-                rt.block_on(async {
-                    run_commands_on_single_thread(limiter, client_config, case, context).await;
-                    wk.done();
-                });
-            });
-            thread_handlers.push(thread_handler);
-        }
-
-        // log thread
-        let result = wait_finish(&case, &conf.client_config(), auto_limiter, context, wg);
-        results.add(result);
-
-        // join all threads
-        for thread_handler in thread_handlers {
-            thread_handler.join().unwrap();
-        }
-    }
-
-    results.save(conf.output_formats);
-}
diff --git a/src/output.rs b/src/output.rs
deleted file mode 100644
index 406f5ba..0000000
--- a/src/output.rs
+++ /dev/null
@@ -1,119 +0,0 @@
-use crate::config::OutputFormat;
-use colored::Colorize;
-use rust_xlsxwriter::{Format, FormatAlign, Workbook};
-use serde::Serialize;
-
-#[derive(Serialize, Default)]
-pub struct Result {
-    pub name: String,
-    pub qps: f64,
-    #[serde(rename = "avg(ms)")]
-    pub avg: f64,
-    #[serde(rename = "min(ms)")]
-    pub min: f64,
-    #[serde(rename = "p50(ms)")]
-    pub p50: f64,
-    #[serde(rename = "p90(ms)")]
-    pub p90: f64,
-    #[serde(rename = "p99(ms)")]
-    pub p99: f64,
-    #[serde(rename = "max(ms)")]
-    pub max: f64,
-    #[serde(rename = "memory(GiB)")]
-    pub memory: f64,
-    // test arguments
-    pub connections: u64,
-    pub pipeline: u64,
-    pub count: u64,
-    #[serde(rename = "duration(s)")]
-    pub duration: f64,
-}
-
-pub(crate) struct Results {
-    results: Vec<Result>,
-}
-
-impl Results {
-    pub fn new() -> Results {
-        Results { results: Vec::new() }
-    }
-
-    pub fn add(&mut self, result: Result) {
-        self.results.push(result);
-    }
-    pub fn save(&self, output_formats: Vec<OutputFormat>) {
-        let filename = std::format!("{}", chrono::Local::now().format("%Y-%m-%d_%H-%M-%S"));
-        for output_format in output_formats {
-            match output_format {
-                OutputFormat::XLSX => {
-                    self.save_xlsx(filename.clone());
-                }
-                OutputFormat::JSON => {
-                    self.save_json("output".to_string());
-                }
-            }
-        }
-    }
-
-    fn save_json(&self, filename: String) {
-        let filename = std::format!("{}.json", filename);
-        let json = serde_json::to_string_pretty(&self.results).unwrap();
-        std::fs::write(&filename, json).unwrap();
-        println!("{} {}", "Saved to".bold().yellow(), filename);
-    }
-
-    fn save_xlsx(&self, filename: String) {
-        let mut workbook = Workbook::new();
-        let worksheet = workbook.add_worksheet();
-        assert!(self.results.len() > 0);
-        let header_format = Format::new().set_bold().set_align(FormatAlign::Center);
-        worksheet.serialize_headers_with_format(0, 0, &self.results[0], &header_format).unwrap();
-        for result in &self.results {
-            worksheet.serialize(result).unwrap();
-        }
-        let float_format = Format::new().set_num_format("0.00").set_align(FormatAlign::Center);
-        let int_format = Format::new().set_num_format("0").set_align(FormatAlign::Center);
-        // command
-        worksheet.set_column_width(0, 50).unwrap();
-        // qps
-        worksheet.set_column_width(1, 15).unwrap();
-        worksheet.set_column_format(1, &int_format).unwrap();
-        // avg
-        worksheet.set_column_width(2, 10).unwrap();
-        worksheet.set_column_format(2, &float_format).unwrap();
-        // min
-        worksheet.set_column_width(3, 10).unwrap();
-        worksheet.set_column_format(3, &float_format).unwrap();
-        // p50
-        worksheet.set_column_width(4, 10).unwrap();
-        worksheet.set_column_format(4, &float_format).unwrap();
-        // p90
-        worksheet.set_column_width(5, 10).unwrap();
-        worksheet.set_column_format(5, &float_format).unwrap();
-        // p99
-        worksheet.set_column_width(6, 10).unwrap();
-        worksheet.set_column_format(6, &float_format).unwrap();
-        // max
-        worksheet.set_column_width(7, 10).unwrap();
-        worksheet.set_column_format(7, &float_format).unwrap();
-        // memory
-        worksheet.set_column_width(8, 15).unwrap();
-        worksheet.set_column_format(8, &float_format).unwrap();
-        // connections
-        worksheet.set_column_width(9, 15).unwrap();
-        worksheet.set_column_format(9, &int_format).unwrap();
-        // pipeline
-        worksheet.set_column_width(10, 15).unwrap();
-        worksheet.set_column_format(10, &int_format).unwrap();
-        // count
-        worksheet.set_column_width(11, 15).unwrap();
-        worksheet.set_column_format(11, &int_format).unwrap();
-        // duration
-        worksheet.set_column_width(12, 15).unwrap();
-        worksheet.set_column_format(12, &float_format).unwrap();
-
-        let filename = std::format!("{}.xlsx", filename);
-        workbook.save(&filename).unwrap();
-        println!("{} {}", "Saved to".bold().yellow(), filename);
-    }
-}
diff --git a/src/shared_context.rs b/src/shared_context.rs
index b7a5a9c..c288382 100644
--- a/src/shared_context.rs
+++ b/src/shared_context.rs
@@ -1,5 +1,5 @@
-use crate::common::async_flag::AsyncFlag;
-use crate::Histogram;
+use crate::async_flag::AsyncFlag;
+use crate::histogram::Histogram;
 use std::cmp::min;
 use std::option::Option;
 use std::sync::atomic::AtomicU64;
diff --git a/workloads/example.toml b/workloads/example.toml
deleted file mode 100644
index 38a755c..0000000
--- a/workloads/example.toml
+++ /dev/null
@@ -1,27 +0,0 @@
-cluster = false
-address = "127.0.0.1:7001"
-username = ""
-password = ""
-tls = false
-db_type = "redis"  # redis, kvrocks, tair_mem, tair_scm, tair_ssd, garnet
-output = ["xlsx", "json"]
-replica_count = 0
-
-# Leave empty to use all available CPUs. You can specify CPU cores,
-# such as "0,1,6-10" to use cores 0, 1, 6, 7, 8, 9, and 10.
-cpus = ""
-
-[[cases]]
-name = "STRING: SET 64B"
-command = "SET {key uniform 50000000} {value 64}"
-connections = 8
-pipeline = 8
-count = 100_000
-
-[[cases]]
-name = "SET: SISMEMBER"
-dataset = { command = "SADD {key sequence 10} {key sequence 107}", count = 1007_000 }
-command = "SISMEMBER {key uniform 10} {key uniform 1007}"
-connections = 0
-pipeline = 1
-seconds = 10
\ No newline at end of file
diff --git a/workloads/redis.toml b/workloads/redis.toml
deleted file mode 100644
index fb06d82..0000000
--- a/workloads/redis.toml
+++ /dev/null
@@ -1,80 +0,0 @@
-cluster = false
-address = "127.0.0.1:6379"
-password = ""
-
-[[cases]]
-name = "NETWORK: ECHO 64B"
-command = "ECHO {value 64}"
-seconds = 120
-
-[[cases]]
-name = "NETWORK: PING with Pipeline"
-command = "PING"
-pipeline = 8
-seconds = 120
-
-[[cases]]
-name = "NETWORK: ECHO 1024B"
-command = "ECHO {value 1024}"
-seconds = 120
-
-[[cases]]
-name = "STRING: SET 64B"
-command = "SET {key uniform 50000000} {value 64}"
-seconds = 120
-
-[[cases]]
-name = "STRING: GET 64B"
-dataset = { command = "SET {key sequence 100000} {value 64}", count = 100000 }
-command = "GET {key uniform 100000}"
-seconds = 120
-
-[[cases]]
-name = "LIST: LPUSH 64B"
-command = "LPUSH {key uniform 10000} {value 64}"
-seconds = 120
-
-[[cases]]
-name = "LIST: LPOP 64B"
-dataset = { command = "LPUSH {key sequence 1000} {value 8}", count = 1007_000 }
-command = "LINDEX {key uniform 100} {rand 1007}"
-seconds = 120
-
-[[cases]]
-name = "SET: SADD"
-command = "SADD {key sequence 100000} {key sequence 1007}"
-seconds = 120
-
-[[cases]]
-name = "SET: SISMEMBER"
-dataset = { command = "SADD {key sequence 1000} {key sequence 1007}", count = 1007_000 }
-command = "SISMEMBER {key uniform 1000} {key uniform 1007}"
-seconds = 120
-
-[[cases]]
-name = "HASH: HSET 64B"
-command = "HSET {key sequence 40000} {key sequence 1007} {value 16}"
-seconds = 120
-
-[[cases]]
-name = "HASH: HGET 64B"
-dataset = { command = "HSET {key sequence 1000} {key sequence 1007} {value 8}", count = 1007_000 } # 1007 elements
-command = "HGET {key uniform 1000} {key uniform 1007}"
-seconds = 120
-
-[[cases]]
-name = "ZSET: ZADD"
-command = "ZADD {key uniform 30000} {rand 6000} {key sequence 1007}"
-seconds = 120
-
-[[cases]]
-name = "ZSET: ZRANGEBYSCORE"
-dataset = { command = "ZADD {key sequence 1000} {rand 1000} {value 8}", count = 1000_000 }
-command = "ZRANGEBYSCORE {key uniform 1000} {range 1000 10}"
-seconds = 120
-
-[[cases]]
-name = "LUA: SET 64B"
-dataset = { command = '''SCRIPT LOAD "return redis.call('SET', KEYS[1], ARGV[1])"''', count = 1 }
-command = "EVALSHA d8f2fad9f8e86a53d2a6ebd960b33c4972cacc37 1 {key uniform 100000} {value 64}"
-seconds = 120