[HIVE-23846] Avoid unnecessary serialization and deserialization of bitvectors

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.0.0-alpha-1
• Component/s: Standalone Metastore
• Labels: pull-request-available
• Assignee: Yu-Wen Lai
• Reporter: Yu-Wen Lai
• Created: 15/Jul/20 01:31
• Updated: 17/Nov/22 08:47
• Resolved: 21/Jul/20 16:50

Description

In the method getNdvEstimator of <datatype>ColumnStatsDataInspector, it will call isSetBitVectors(), in which it serializes the bitvectors again when we already have deserialized bitvectors ndvEstimator. For example, we can see this pattern from LongColumnStatsDataInspector.

This method could check if the ndvEstimator is set first so that it won't need to serialize and deserialize back.

Issue Links

• is related to: HIVE-17286 Avoid expensive String serialization/deserialization for bitvectors
• links to: GitHub Pull Request #1266