[ 
https://issues.apache.org/jira/browse/CASSANDRA-2558?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13025131#comment-13025131
 ] 

Sylvain Lebresne commented on CASSANDRA-2558:
---------------------------------------------

To be clear:
  # I wouldn't at all be surprised that Terje report was an experiment and not 
a real life situation (that's why I asked, he may confirm that), but
  # I do not believe in "It works fine for us, why wouldn't it work for 
others". And more precisely
  # I'm more worried about cases like: throttling is not very high but there is 
a temporary higher load of writes. New compactions may be started quicker than 
it task for ones to finish (and the more compaction, the less quickly each one 
goes with throttling). It may then be better to limit the number of compactions 
rather than the alternative.
  # Adding this option is trivial so whether my preceding points make sense or 
not I'm not sure it's worth debating adding the option. What could be "debated" 
is whether we should "expose" it although adding it as a commented option with 
same default as today don't cost us anything. And honestly, if "in practice, 
you won't see more than 2-3 compactions", what about limiting it to 4 or 6 by 
default as a safety net ? 

> 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

Reply via email to