[jira] [Created] (CASSANDRA-11885) Profile current compaction performance

["Carl Yeksigian (JIRA)" <ji...@apache.org>]

Carl Yeksigian created CASSANDRA-11885:
------------------------------------------

             Summary: Profile current compaction performance
                 Key: CASSANDRA-11885
                 URL: https://issues.apache.org/jira/browse/CASSANDRA-11885
             Project: Cassandra
          Issue Type: Sub-task
          Components: Compaction
            Reporter: Carl Yeksigian
            Assignee: Carl Yeksigian


With CASSANDRA-11844, we can create compaction stress profiles. We need to get a baseline for the current compaction work so that we can gauge future improvements.

Scenarios to test:
- Compacting non-overlapping sstables
- Compacting sstables with non-overlapping partitions, but mixed together in merge order
- Compacting wide rows, but non-overlapping sstables
- Compacting wide rows, with no cell merges between sstables
- Compacting wide rows, with cell merges between sstables

Each tests a different workload, so we need to make sure that whatever performance improvements we make don't adversely affect the other workloads. We should vary the number of sstables that we compact to get a spectrum of the throughput performance. We should also make this into a python script so that it can be rerun whenever we need to.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)