What's New in Apache Kudu

Learn about the new features of Kudu in Cloudera Runtime 7.2.18.

Auto-incrementing column

Kudu now supports backup and restore of tables with auto incrementing columns. The restored table will have the auto incrementing column values identical to the source table for every row. For more details, see Non-unique primary key index.

Kudu Range-aware Data Placement

Kudu places new tablet replicas using an algorithm which is both range and table aware. This algorithm helps to avoid hotspotting that occurs if many replicas backing the same range are placed on the same few tablet servers. Hotspotting causes tablet servers to be overwhelmed with write or read requests and can result in increased latency for these requests. To avoid hotspotting, this algorithm avoids targeting the same set of tablet servers for a set of replicas created in parallel. Rather, it spreads the replicas across multiple tablet servers. For more information, see How Range-aware replica placement in Kudu works.

Kudu multi-master config change

You can now remove or decommission the unwanted master role instances through Cloudera Manager. Also, you can recommission any decommissioned master role instance in a multi-master deployment. For more information, see Removing Kudu masters through Cloudera Manager.

Improvements

None.