TDengine
1. Introduction
TDengine is an open source, high-performance, cloud native and AI powered time-series database designed for Internet of Things (IoT), Connected Cars, and Industrial IoT. It enables efficient, real-time data ingestion, processing, and analysis of TB and even PB scale data per day, generated by billions of sensors and data collectors. TDengine differentiates itself from other time-series databases with the following advantages:
High Performance: TDengine is the only time-series database to solve the high cardinality issue to support billions of data collection points while out performing other time-series databases for data ingestion, querying and data compression.
Simplified Solution: Through built-in caching, stream processing, data subscription and AI agent features, TDengine provides a simplified solution for time-series data processing. It reduces system design complexity and operation costs significantly.
Cloud Native: Through native distributed design, sharding and partitioning, separation of compute and storage, RAFT, support for kubernetes deployment and full observability, TDengine is a cloud native Time-Series Database and can be deployed on public, private or hybrid clouds.
AI Powered: Through the built in AI agent TDgpt, TDengine can connect to a variety of time series foundation model, large language model, machine learning and traditional algorithms to provide time series data forecasting, anomaly detection, imputation and classification.
Ease of Use: For administrators, TDengine significantly reduces the effort to deploy and maintain. For developers, it provides a simple interface, simplified solution and seamless integrations for third party tools. For data users, it gives easy data access.
Easy Data Analytics: Through super tables, storage and compute separation, data partitioning by time interval, pre-computation and AI agent, TDengine makes it easy to explore, format, and get access to data in a highly efficient way.
Open Source: TDengine's core modules, including cluster feature and AI agent, are all available under open source licenses. It has gathered 23.7k stars on GitHub. There is an active developer community, and over 730k running instances worldwide.
For a full list of TDengine competitive advantages, please check here. The easiest way to experience TDengine is through TDengine Cloud. For the latest TDengine component TDgpt, please refer to TDgpt README for details.
2. Documentation
For user manual, system design and architecture, please refer to TDengine Documentation (TDengine 文档)
You can choose to install TDengine via container, installation package, Kubernetes or try fully managed service without installation. This quick guide is for developers who want to contribute, build, release and test TDengine by themselves.
Table of Contents
3. Prerequisites
3.1 System Requirements
- Operating systems: Linux (Ubuntu 18.04+, CentOS 7+), macOS 10.15+, Windows (limited; open-source builds are primarily Linux/macOS)
- CPU: x86_64 or ARM64
- Memory: 4 GB RAM or more recommended
- Disk: 2 GB free space or more recommended
- Primary build platform: Linux
TDengine is built and tested primarily on Linux. macOS builds are supported for local development. Windows support in the open-source tree is limited, so Linux should be your default choice for reproducible builds.
Ubuntu/Debian:
sudo apt-get update
sudo apt-get install -y build-essential cmake git
CentOS/RHEL:
sudo yum groupinstall -y "Development Tools"
sudo yum install -y cmake3 git
Note: CMake >= 3.21 is required for external builds. If your distro provides an older version, install a newer release from https://cmake.org/download/.
You may also need these tools depending on what you build:
- Python 3 for the test framework in
tests/
- Go 1.23+ if you want to build components such as taosAdapter or taosKeeper
- ccache — speeds up rebuilds:
sudo apt install ccache
- Conan 2.x — required only for the
taos-gen component: pip3 install conan
4. Building
4.1 Quick Start
git clone https://github.com/taosdata/TDengine.git
cd TDengine
mkdir debug && cd debug
cmake .. -DBUILD_CONTRIB=ON # First build: -DBUILD_CONTRIB=ON is REQUIRED
make -j$(nproc)
Important
The first source build must use -DBUILD_CONTRIB=ON. This downloads and builds external dependencies such as xxhash, zstd, and lz4 into .externals/. Subsequent builds can usually omit this flag and reuse the cached artifacts automatically.
Useful variants:
# Release build with CLI tools
cmake .. -DBUILD_CONTRIB=ON -DCMAKE_BUILD_TYPE=Release -DBUILD_TOOLS=ON
make -j$(nproc)
# macOS equivalent parallel build
cmake .. -DBUILD_CONTRIB=ON -DCMAKE_BUILD_TYPE=Release
cmake --build . --parallel "$(sysctl -n hw.ncpu)"
4.2 Build Options
Defaults below are the standalone open-source defaults from cmake/options.cmake unless noted otherwise.
| Option |
Default |
Description |
BUILD_CONTRIB |
OFF on Linux, ON elsewhere |
Build third-party dependencies from source. Required for the first build. |
CMAKE_BUILD_TYPE |
Debug |
Standard CMake build type (Debug, Release, RelWithDebInfo). |
BUILD_TEST |
OFF |
Build unit tests with googletest. |
BUILD_TOOLS |
OFF |
Build tools such as taosBenchmark and taosdump. |
BUILD_SANITIZER |
OFF |
Enable sanitizers. |
BUILD_COVERAGE |
OFF |
Enable coverage instrumentation. |
BUILD_JEMALLOC |
OFF |
Enable jemalloc allocator support. |
BUILD_WEBSOCKET |
OFF |
Enable WebSocket support. |
BUILD_ASSERT_NOT_CORE |
OFF |
Avoid generating core files on assert. |
BUILD_PTHREAD_TWEAK |
OFF |
Apply pthread tweaks for older Linux environments. |
BUILD_ASTRA |
OFF |
Build for Astra platform. |
BUILD_ASTRA_RPC |
OFF |
Build with Astra RPC transport. |
BUILD_WITH_LEMON |
ON |
Build with lemon parser support. |
BUILD_WITH_UDF |
ON |
Build with UDF support. |
BUILD_GEOS |
ON |
Build with GEOS support (non-Astra). |
BUILD_SHARED_LIBS |
OFF |
Build shared libraries. |
RUST_BINDINGS |
ON |
Build Rust bindings. |
BUILD_PCRE2 |
ON |
Build with PCRE2 support. |
BUILD_ADDR2LINE |
OFF |
Build addr2line helper support. |
BUILD_WITH_LEVELDB |
OFF |
Enable LevelDB support. |
BUILD_ROCKSDB |
OFF on Linux, ON elsewhere |
Build RocksDB from source. |
ROCKSDB_USE_DEPS |
ON on Linux, OFF elsewhere |
Use prebuilt RocksDB from deps/ instead of building it. |
TD_USE_ROCKSDB |
ON |
Enable RocksDB support. |
BUILD_WITH_LZ4 |
ON |
Build with LZ4 support. |
BUILD_S3 |
ON on Linux, then forced OFF for community builds |
Enable S3-related build path. Community builds switch this off. |
BUILD_WITH_S3 |
ON on Linux, then forced OFF for community builds |
Build with S3 support when available. |
BUILD_WITH_COS |
OFF |
Build with COS support. |
BUILD_WITH_LZMA2 |
ON |
Build with LZMA2 support. |
BUILD_WITH_ANALYSIS |
ON on Linux |
Enable analysis-related build path. |
BUILD_WITH_SQLITE |
OFF |
Build with SQLite support. |
BUILD_WITH_BDB |
OFF |
Build with Berkeley DB support. |
BUILD_WITH_LUCENE |
OFF |
Build with Lucene support. |
BUILD_WITH_NURAFT |
OFF |
Build with NuRaft support. |
BUILD_WITH_UV |
ON |
Build with libuv support. |
BUILD_WITH_UV_TRANS |
ON |
Build with libuv transport support. |
BUILD_DEPENDENCY_TESTS |
ON on Linux |
Build dependency tests. |
BUILD_DOCS |
OFF |
Build Doxygen documentation. |
BUILD_WITH_INVERTEDINDEX |
ON |
Enable inverted index support. |
BUILD_TAOSD_INTEGRATED |
OFF |
Build taosd as an integrated library. |
BUILD_AS_LIB |
OFF |
Build TDengine as a library. |
BUILD_RELEASE |
OFF |
Enable release-version build path. |
BUILD_LIBSASL |
OFF |
Build with libsasl2. |
BUILD_FLEX_DEPLOY |
OFF |
Enable flexible deployment mode. |
BUILD_WITH_RAND_ERR |
OFF |
Enable random error injection. |
BUILD_TSZ_ENABLED |
ON |
Enable TSZ compression support. |
BUILD_USE_PUBLIC_DEPS |
OFF |
Use public internet URLs for external dependencies instead of alternate mirrors. |
Example:
cmake .. -DBUILD_CONTRIB=ON -DCMAKE_BUILD_TYPE=Release -DBUILD_TOOLS=ON
4.3 Build Output
The default build tree under debug/ looks like this:
- Binaries in
debug/build/bin/
taosd — TDengine server daemontaos — TDengine CLI clienttaosBenchmark — benchmark tool (BUILD_TOOLS=ON)taosdump — import/export tool (BUILD_TOOLS=ON)
- Libraries in
debug/build/lib/
libtaos.so on Linux (or the platform-equivalent client library)
5. Testing
5.1 Unit Tests
cd debug
cmake .. -DBUILD_CONTRIB=ON -DBUILD_TEST=ON
make -j$(nproc)
ctest --output-on-failure
Some individual test binaries are also emitted into debug/build/bin/.
5.2 Integration Tests
TDengine includes a Python-based test framework in tests/.
# Build the server and tools first
cd debug
cmake .. -DBUILD_CONTRIB=ON -DBUILD_TOOLS=ON -DBUILD_TEST=ON
make -j$(nproc)
# Start taosd with the generated test configuration
./build/bin/taosd -c test/cfg &
# Run a system test
cd ../tests/system-test
python3 test.py -f 2-query/basic.py
For more test details, see tests/README.md.
6. Packaging
Use source/taos-community/packaging/pack_community_tar.sh after a successful Linux build to assemble the open-source community tarballs.
This script packages only artifacts that are built from the taos-community repository itself:
taosdtaostaosBenchmark (BUILD_TOOLS=ON, when present)taosdump (BUILD_TOOLS=ON, when present)taosudf (when present)libtaos.solibtaosnative.so- headers,
examples/c, and the install scripts
It does not package components that live in other repositories or require separate build flows, such as taoskeeper, taos-explorer, taosx, connectors, or taosinspect.
Build first:
mkdir debug && cd debug
cmake .. -DCMAKE_BUILD_TYPE=Release -DBUILD_CONTRIB=ON -DBUILD_TOOLS=ON
make -j$(nproc)
Then run the packaging script from the repository root:
cd ..
./source/taos-community/packaging/pack_community_tar.sh -c debug -n 3.3.6.0
Optional arguments:
-m <compatible_version> — defaults to 3.0.0.0-V <stable|beta|preRelease> — defaults to stable
Example:
./source/taos-community/packaging/pack_community_tar.sh \
-c debug \
-n 3.3.6.0 \
-m 3.0.0.0 \
-V stable
Generated packages are written to:
source/taos-community/release/
Typical outputs:
source/taos-community/release/TDengine-server-3.3.6.0-Linux-x64.tar.gz
source/taos-community/release/TDengine-client-3.3.6.0-Linux-x64.tar.gz
The server and client archives use the same two-layer layout as the existing packaging flow:
- outer tarball for distribution
- inner
package.tar.gz containing bin/, cfg/, and inc/
This is the standard packaging approach for repository-local OSS artifacts and should be followed consistently in future repositories as well.
7. Installing from Package
For an official installer package produced by packaging/tools/makepkg.sh:
tar -xzf TDengine-server-<version>-Linux-x86_64.tar.gz
cd TDengine-server-<version>
sudo ./install.sh
If you are installing directly from a local source build instead of a packaged archive, you can also run:
cd debug
sudo make install
8. Running
# Start the server after installation
sudo systemctl start taosd
# Connect with the CLI
taos
For a quick local run from the build tree, use:
cd debug
./build/bin/taosd -c test/cfg
Then, in another shell:
cd debug
./build/bin/taos -c test/cfg
9. Workflow
TDengine build check workflow can be found in this GitHub Action. More workflows will be available soon.
10. Coverage
Latest TDengine test coverage report can be found on coveralls.io.
How to run the coverage report locally?
To create the test coverage report (in HTML format) locally, please run following commands:
cd tests
bash setup-lcov.sh -v 1.16 && ./run_local_coverage.sh -b main -c task
# on main branch and run cases in longtimeruning_cases.task
# for more information about options please refer to ./run_local_coverage.sh -h
NOTE:
Please note that the -b and -i options will recompile TDengine with the -DCOVER=true option, which may take an amount of time.
11. Contributing
Please follow the contribution guidelines to contribute to TDengine.
12. License
TDengine is licensed under the GNU Affero General Public License 3.0.
TDengine 是一款开源、高性能、云原生、支持 AI 的时序数据库，专为物联网、车联网和工业物联网场景设计，可处理从 TB 到 PB 级、来自数十亿传感器的数据。