TDengine — High-Performance Time-Series Database for IoT

What it is

TDengine is a purpose-built time-series database optimized for IoT, industrial telemetry, and monitoring data. It combines a storage engine designed for time-series patterns with built-in caching, stream processing, and data subscription in a single platform. The SQL-compatible interface means teams can adopt it without learning a new query language.

TDengine targets IoT platform teams, DevOps engineers collecting metrics, and any organization dealing with high-volume time-stamped data that outgrows general-purpose databases.

How it saves time or tokens

General-purpose databases like PostgreSQL or MySQL waste storage on time-series data because they lack column-oriented compression tuned for sequential timestamps. TDengine's columnar storage with domain-specific compression achieves up to 10x better compression ratios. Queries on time ranges are also faster because the engine skips irrelevant blocks at the storage level. Built-in stream processing eliminates the need for separate Kafka or Flink pipelines for basic aggregations.

How to use

1. Start TDengine with Docker:

docker run -d -p 6030:6030 -p 6041:6041 tdengine/tdengine

1. Connect via the CLI client:

taos

1. Create a database and insert data:

CREATE DATABASE demo;
USE demo;
CREATE STABLE meters (ts TIMESTAMP, voltage FLOAT, current FLOAT) TAGS (location VARCHAR(64), group_id INT);
CREATE TABLE meter_01 USING meters TAGS ('factory_a', 1);
INSERT INTO meter_01 VALUES (NOW, 220.5, 1.2);
SELECT AVG(voltage) FROM meters WHERE ts > NOW - 1h GROUP BY location;

Example

Using TDengine's super-table pattern for IoT devices:

-- Create a super-table (schema template)
CREATE STABLE sensors (
  ts TIMESTAMP,
  temperature FLOAT,
  humidity FLOAT
) TAGS (device_id VARCHAR(32), floor INT);

-- Auto-create sub-tables per device
INSERT INTO d_abc USING sensors TAGS ('abc', 3)
  VALUES (NOW, 23.5, 45.2);

-- Query across all devices
SELECT device_id, AVG(temperature)
FROM sensors
WHERE ts > NOW - 24h
GROUP BY device_id;

Related on TokRepo

• AI Tools for Database — database tools and management utilities
• AI Tools for Monitoring — monitoring and observability platforms

Common pitfalls

• TDengine's super-table model requires planning your tag structure upfront; poorly chosen tags lead to inefficient queries
• The REST API (port 6041) and native protocol (port 6030) serve different use cases; use the native protocol for high-throughput ingestion
• Cluster mode requires at least 3 nodes for data replication; single-node deployments cannot be non-disruptively expanded to clusters