What’s New in openLooKeng 1.6.0? Key Performance Boosts and Features

openLooKeng v1.6.0 Release

openLooKeng 1.6.0 brings a series of optimizations and new features aimed at improving engine performance, reliability, and functionality.

1. Data Persistence

Optimized calculation data spill speed and size by using Kryo serialization, which reduces write time and file size.

Added support for spilling overflow data to HDFS, increasing throughput.

Implemented an asynchronous spill/no‑spill mechanism to avoid blocking downstream operators when memory thresholds are exceeded.

Enabled overflow write for right‑outer‑join and full‑join scenarios using a Bloom filter to identify overflow data.

2. Connector Enhancements

PostgreSQL and openGauss connectors now support update and delete operations on the data source.

3. Low Latency

Added Star Tree index support, cube update commands, and heuristic index optimizations that shrink Bloom filter size by more than ten times.

4. Task Recovery

Reduced task failure detection time from 300 seconds to a much shorter interval, improving overall query execution flow.

Optimized snapshot time and size by replacing Java serialization with Kryo, resulting in faster snapshots and smaller files.

Note: Kryo is a fast serialization/deserialization tool that leverages bytecode generation (using the ASM library) to achieve high performance.

openLooKeng maintains a three‑month release cadence, continuing to provide a fast, minimalist data experience. Download links and a QR code for the new version are available below.