-
Type: Task
-
Resolution: Done
-
Priority: Major - P3
-
None
-
Affects Version/s: None
-
Component/s: Performance, Replication
-
Replication
Most of the performance work on oplog shipping and application has been focused on improving throughput when the servers are heavily loaded with writes. The goal of this ticket is to examine and possibly improve the latency of w>1 operations (or equivalently, the throughput of single-threaded operations). An area of concern is waiting for more data before processing a batch.