[
https://issues.apache.org/jira/browse/CASSANDRA-9533?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14637765#comment-14637765
]
Brandon Williams commented on CASSANDRA-9533:
---------------------------------------------
I think we need something sooner than 3.0 for this, because single-threaded
performance is still worse than 2.0 with a 1ms batch:
http://cstar.datastax.com/graph?stats=c8ca39e6-30ad-11e5-aceb-42010af0688f&metric=op_rate&operation=1_write&smoothing=1&show_aggregates=true&xmin=0&xmax=450.89&ymin=0&ymax=31.9
With stress defaults they have parity, but 2.1 loses when you switch to a 30k
payload, and gets worse with a 64k payload:
http://cstar.datastax.com/graph?stats=7c92f43c-30bc-11e5-ac71-42010af0688f&metric=op_rate&operation=1_write&smoothing=1&show_aggregates=true&xmin=0&xmax=544.5&ymin=0&ymax=27.5
> Make batch commitlog mode easier to tune
> ----------------------------------------
>
> Key: CASSANDRA-9533
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9533
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Jonathan Ellis
> Assignee: Benedict
> Fix For: 3.x
>
>
> As discussed in CASSANDRA-9504, 2.1 changed commitlog_sync_batch_window_in_ms
> from a maximum time to wait between fsync to the minimum time, so one must be
> very careful to keep it small enough that most writers aren't kept waiting.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
