[ARROW-5482] [Java] reduce heap footprint of ValueVectors

• Type: Task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Java
• External issue URL: https://github.com/apache/arrow/issues/21931
• Assignee: Unassigned
• Reporter: Pindikura Ravindra
• Created: 03/Jun/19 05:41
• Updated: 11/Jan/23 07:40

Description

In some scenarios, we hold lots of value vectors in memory eg. during join, aggregation. The heap analysis shows that the costs are as follows for a simple IntVector (used VisualVM on mac) :

 

IntVector : 80 bytes

vector.types.pojo.FieldType : 41 bytes

vector.types.pojo.Field : 40 bytes

IntReaderImpl : 28 bytes

 

I'll use this Jira to track ways to reduce the heap usage.

 

Issue Links

• is a parent of: ARROW-5484 [Java] remove FieldReader from ValueVector
• is a parent of: ARROW-5483 [Java] add ValueVector constructors that take a Field object