LoginSign Up
← Back to all sparks
A

Apache IoTDB

DEVOPS
Velocity2.5

Time series database for IoT and industrial data

iotdb.apache.org ↗

Apache IoTDB is closing the SQL expressiveness gap while keeping its IoT-native core.

time-seriesiot-databasesql-parityembedded-analyticsapache
◆Current state
IoTDB 2.x has reached a level of SQL completeness — set operations, CTEs, window functions, JOIN variants, MATCH RECOGNIZE, and now logical views — that makes it viable for data engineers who previously had to export time-series data into a relational database for complex analysis. The 1.3.x branch is in maintenance mode, receiving only security backports. The AINode capability adds built-in ML models (Timer-XL, Timer-Sundial) for in-database forecasting.
◆Where it's heading
The 2.x line is systematically adding relational SQL expressiveness atop the IoT-native storage core, adding 2-4 SQL features per release. The C-language SDK signals an intent to expand beyond JVM-centric deployments into embedded and industrial control contexts. AINode points toward a longer arc: time-series forecasting and anomaly detection executed directly in the database, reducing the need to export data to Python for ML workflows.
◆Prediction
The next releases will likely complete table model SQL parity with standard features still missing, and expand AINode inference to cover more model types or expose forecasting via standard SQL function syntax.

◆Recent moves

  1. 5d ago

    IoTDB 2.0.11: logical views, JDK 17 required, EXPLAIN ANALYZE JSON output

    Logical views complete a key relational abstraction — users can now define query-time views over physical time-series data without materializing them. The JDK 17 minimum is a deployment breaking change operators must plan for. EXPLAIN ANALYZE now outputs JSON, improving integration with external query tooling.

    View source ↗
  2. 2mo ago

    IoTDB 2.0.10: set operations, CTEs, and a C-language SDK

    UNION, INTERSECT, EXCEPT, and Common Table Expressions bring the table model to SQL standard parity for analysts running multi-source queries. The C-language SDK with automatic multi-node failover opens IoTDB to embedded C/C++ workloads in industrial and control environments — a meaningful surface expansion beyond the JVM ecosystem.

    View source ↗
  3. 5mo ago

    IoTDB 2.0.8: Python DataFrame support and query latency observability

    The Python SDK now converts query results directly to pandas DataFrames in batches, cutting boilerplate for data science workflows. A new system table for query latency statistics adds observability that previously required external tooling. Both changes lower the barrier for Python-native data engineers working with IoTDB.

    View source ↗
  4. 6mo ago

    IoTDB 2.0.7: RPC surface reduction and default address hardening

    Security hardening release: risky RPC interfaces and JEXL scripting removed, default client address changed from open binding to 127.0.0.1, and internal services bound to the configured address only. One partition table TTL bug fixed. No user-visible features added.

    View source ↗
  5. 6mo ago

    IoTDB 1.3.7: security hardening backport to maintenance branch

    The same security changes from 2.0.7 — RPC removal, JEXL removal, address hardening — backported to the 1.3 maintenance line. This confirms 1.3.x receives only critical fixes while development investment goes into 2.x.

    View source ↗
  6. 7mo ago

    IoTDB 2.0.6: MATCH RECOGNIZE for event detection, query write-back, CVE fixes

    MATCH RECOGNIZE brings native row pattern recognition to the table model, enabling detection of continuous data patterns — state transitions, anomaly sequences — directly in SQL without post-processing. Query write-back functions complete the read-modify-write loop in SQL. Three CVEs patched. MATCH RECOGNIZE in particular is a capability with few SQL equivalents in other time-series databases.

    View source ↗