At a Glance:
TDengine is an open-source, high-performance time-series database built for IoT and Industrial IoT, combining native distributed cloud design, built-in caching and stream processing, and an AI agent for data forecasting and anomaly detection.
Overview:
TDengine is an open-source, cloud native time-series database engineered for large-scale IoT, Connected Car, and Industrial IoT workloads. It addresses high-cardinality data ingestion at scale, handling billions of data collection points while maintaining high performance for writes, queries, and compression. The platform integrates caching, stream processing, data subscription, and an AI agent directly into the database, reducing the need for additional processing components. Deployment spans public, private, and hybrid clouds with Kubernetes support, while a command-line interface, client library, and structured packaging provide access paths tailored to administrators, developers, and data users managing petabyte-scale time-series data.
Key Decision Points:
Database engine: A time-series database specifically designed for high-cardinality IoT and Industrial IoT data, prioritizing ingestion performance and compression at extreme scale.
Built-in processing: Caching, stream processing, and data subscription run inside the database, lowering external system dependencies for real-time workloads.
AI integration: A built-in AI agent connects to time-series foundation models, large language models, and ML algorithms for forecasting, anomaly detection, imputation, and classification.
Deployment model: Supports native distributed architecture, Kubernetes, and separation of compute and storage, targeting public, private, and hybrid cloud environments.
Access interfaces: Provides a CLI client (
taos), a client library (libtaos), and an open-source community packaging flow for server and client artifacts.Build and packaging scope: Community packaging only includes core artifacts such as
taosd,taos,taosBenchmark,taosdump, and libraries, while other components like connectors require separate build flows.
Core Features:
High-cardinality data support: Engine capable of handling billions of data collection points without performance degradation.
Built-in stream processing: Processes time-series data in real time without requiring external stream processing engines.
Data subscription service: Supports continuous data subscription to push data changes to downstream consumers.
AI agent (TDgpt): Built-in AI component connecting to foundation models, LLMs, and traditional algorithms for time-series analytics tasks.
Cloud native architecture: Natively supports sharding, partitioning, compute-storage separation, and Kubernetes deployment.
Super tables: A data model abstraction enabling efficient data exploration and formatting across partitioned time-series datasets.
Use Cases:
IoT platform operators ingesting and querying data from millions to billions of sensors at high frequency.
Industrial IoT teams requiring real-time stream processing and data subscription without deploying additional processing infrastructure.
Data analysts performing forecasting and anomaly detection on time-series data using TDengine's built-in AI agent.
Open-Source Alternative Value:
TDengine provides its core modules, including clustering and AI agent capabilities, under open-source licenses. Users can self-build the database server, client, and tools from the community repository, run instances locally or on their own infrastructure, and inspect the engine's high-cardinality handling and real-time processing design. The project maintains an active developer community with over 23,000 GitHub stars, offering developers direct access to a production-scale time-series database architecture without dependency on proprietary streaming or analytics platforms.