-
Type:
Sub-task
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
-
Cluster Scalability
-
Cluster Scalability Priorities
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Chunk migration currently uses average document size to find the split points when calling splitVector. This new feature is going to decide the split point based on the actual document size (or better approximation than average collection doc size).
One possible optimization is to start recording the recordIds instead of rescanning them again during cloning (potentially separate ticket if too much work).
- related to
-
SERVER-138891 Integrate balancer logic to handle irregularly sized chunks
-
- Backlog
-