[ORC-777] Make the vectorized row batch size configurable in MR record readers and writers

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.7.0, 1.6.7
• Fix Version/s: 1.7.0, 1.6.8
• Component/s: Java, MapReduce
• Assignee: Sudarshan Vasudevan
• Reporter: Sudarshan Vasudevan
• Created: 29/Mar/21 15:53
• Updated: 21/Sep/21 20:45
• Resolved: 08/Apr/21 09:39

Description

Current behavior of OrcInput(Output)Format is to create an ORC reader/writer with the default row batch size of 1024. This may need to be overridden in certain scenarios to prevent OOMs.

Issue Links

• links to: GitHub Pull Request #674