Apache Kylin 3.0 Released, with greatly improved realtime OLAP features

December 20, Apache Kylin development team announced the latest major release of Apache Kylin 3.0.0.  Many improvements focus on realtime OLAP features.

Apache Kylin is an Online Analytical Processing (OLAP) engine.  Aiming to be “the OLAP in the big data era,” by utilizing multi-dementional and recalculation technology on Apache Hadoop and Spark, it achieves the near constant query speed no matter how large the data volume may be.

As of a new feature, now column row count can be merged.  Also, cube auto merge feature is improved.  Before the latest release, a trigger used to be an event of a new segment ready to be automated, but there was a problem of having too many merge jobs.  However, now in 3.0, there’s a REST API for merging, providing greater manageability.

Other notable changes are that scheduler now supports safe mode, and dimension dictionary can be built using Spark.  Realtime OLAP is greatly improved, and streaming coordination feature has been refactored.  Thanks to these feature improvements, the query latency is reduced to sub-seconds during data streaming.  Now users can set the cube metadata authentication rules.  There are also numerous bug fixes.

Apache Kylin 3 is available on the project’s website.

Apache Kylin