[ https://issues.apache.org/jira/browse/CASSANDRA-2558?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13025126#comment-13025126 ]
Stu Hood commented on CASSANDRA-2558: ------------------------------------- We've been running with multithreaded compaction for a few months now, and in practice, you won't see more than 2-3 compactions running in parallel unless you recently changed your thresholds or triggered something manually (including repair). I'm curious as to how Terje managed to get 24 compactions running at once, but IMO the compaction thresholds are a sufficient tunable to bound the number of parallel tasks. The amount of truly random IO is fairly limited given that our input buffers are 1MB, although our output buffer size recently shrank to 64KB. > Add "concurrent_compactions" configuration > ------------------------------------------ > > Key: CASSANDRA-2558 > URL: https://issues.apache.org/jira/browse/CASSANDRA-2558 > Project: Cassandra > Issue Type: Improvement > Components: Core > Affects Versions: 0.8 beta 1 > Reporter: Sylvain Lebresne > Assignee: Sylvain Lebresne > Priority: Trivial > Fix For: 0.8.0 > > Original Estimate: 2h > Remaining Estimate: 2h > > We should expose a way to configure the max number of thread to use when > multi_threaded compaction is turned on. So far, it uses nb_of_processors > thread, which if you have many cores may be unreasonably high (as far as > random IO is concerned and thus independently of compaction throttling)... at > least unless you have SSD. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira