How Apache IoTDB Dominated Benchmarks and Powered Industry 2023‑2025

Background and Milestones

Apache IoTDB originated in 2011, entered the Apache Incubator in 2018, and became the first top‑level Apache project from a Chinese university. In December 2023 a stack‑style time‑series management architecture and an internal machine‑learning module (AI Node) were released. Since then four versions (1.0‑1.3) have been delivered, with more than 120 developers contributing roughly 950 000 lines of code, adding 85 new features, and validated by about 10 000 open‑source users.

Modular Edge‑Cloud Architecture

Version 1.3 reorganized the system into three interchangeable components built on the proprietary file format TsFile :

IoT Node – lightweight embedded storage for edge devices.

Data Node – high‑performance engine for cloud‑side ingestion, query and compression.

AI Node – in‑database Python runtime that executes data‑quality checks, forecasting and missing‑value imputation without exporting data.

TsFile, originally Java‑only, now provides native libraries for C, C++ and Python and runs on Linux as well as real‑time operating systems such as SylixOS. A benchmark on SylixOS shows 2 MB memory consumption while writing 2 million points per second with more than 10× lossless compression.

Engine Enhancements

Internal stream‑processing operators, a publish/subscribe mechanism between instances, and a multi‑dimensional real‑time processing framework enable chaining of multiple IoTDB instances (pipeline, parallel or hybrid topologies). This allows arbitrary data composition, flow control and dynamic transformation across nodes.

AI Node Capabilities

The AI Node embeds a Python interpreter. Users can submit python scripts that operate directly on TsFile data, e.g.,

# Example: simple moving‑average forecast inside AI Node
import pandas as pd
df = read_tsfile("sensor.ts")
df['ma'] = df['value'].rolling(window=10).mean()
write_tsfile(df, "sensor_ma.ts")

Such scripts run in the same process as the storage engine, eliminating data‑movement overhead.

Benchmark Leadership

IoTDB has held the #1 position on the neutral benchANT TSBS benchmark for more than 20 months, outperforming competitors in write latency, query latency and compression ratio. In August 2024 IoTDB set a new world record on the TPCx‑IoT benchmark, improving performance and cost by roughly 50 % over the previous leader (AMD) on a fully domestic hardware stack.

Community Growth

GitHub stars increased by about 50 % and the number of domestic user groups grew 72 % year‑over‑year. Binary downloads exceed 650 000; Maven and PyPI client libraries have been downloaded over 7 million times, with overseas users accounting for more than 50 % of traffic.

Industrial Deployments

IoTDB is used in more than 1 000 enterprise customers across energy, steel, aerospace and oil‑gas. Representative cases include:

Rocket telemetry lifecycle management for a commercial space company.

On‑board and inter‑satellite data handling on BeiDou‑2/3 satellites.

National pipeline monitoring system replacing the PI System, covering dozens of pipelines and hundreds of stations.

Off‑shore platform time‑series service for Southeast Asian oil rigs, delivering an estimated £1.8 million annual economic benefit.

Future Direction – IoTDB 2.0

IoTDB 2.0 will extend the DB+AI convergence by adding tighter integration of AI workloads, support for additional edge runtimes, and further modularization of the edge‑cloud stack, aiming to meet emerging requirements of digital transformation.