[
https://issues.apache.org/jira/browse/CASSANDRA-9896?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14644042#comment-14644042
]
Benedict commented on CASSANDRA-9896:
-------------------------------------
For very low concurrency (threads=1) and low latency disks (SSDs) "nosleep" is
certainly going to be faster. Is this a workload we consider a priority though?
I suspect the distance disappears on almost all machines once you hit
concurrency of 128 (or less).
CASSANDRA-9533 is the ticket where this particular aspect will be addressed. It
just hasn't been prioritized as there's so much else on our plates right now.
It's not that it's particularly challenging (although we will avoid mutexes in
our implementation)
> Add ability to disable commitlog recycling
> ------------------------------------------
>
> Key: CASSANDRA-9896
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9896
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Brandon Williams
> Assignee: Benedict
> Fix For: 2.1.x
>
>
> See CASSANDRA-9533 for background, specifically the graphs I linked.
> Benedict suggests this is due the commitlog recycling and I agree, so the
> simplest solution is to be able to disable that.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
