Details
-
Sub-task
-
Status: Open
-
Major
-
Resolution: Unresolved
-
0.11.0.0
-
None
-
None
-
None
Description
Currently, we don't have an idea of the throughput curve for transactions across a different range of workloads.
Thus we would like to understand how to tune transactions so that they are viable across a broad range of work loads. For instance, what knobs can you tweak if you use small messages to yet get acceptable transactional performance? We don't understand the performance curve across variables like message size, batch size, transaction duration, linger.ms, etc., and it would be good to get an understanding of this area and publish recommended configurations for different workloads.