[ https://issues.apache.org/jira/browse/CASSANDRA-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14299074#comment-14299074 ]
T Jake Luciani edited comment on CASSANDRA-8692 at 1/30/15 7:12 PM: -------------------------------------------------------------------- Running some experiments I can see a big improvement with this ticket on a 6 node RF=3 cluster writing at QUORUM. The left half is without the patch the right half is with. From steady state of 80k/sec to a steady state of 120k/sec I will mention I get similar results by using nagle. However just enabling nagle will cause small requests to slow down where as this approach adjusts. !batching-benchmark.png! was (Author: tjake): Running some experiments I can see a big improvement with this ticket on a 6 node RF=3 cluster writing at QUORUM. The left half is without the patch the right half is with. From steady state of 80k/sec to a steady state of 120k/sec I will mention I get similar results by using nagle. However just enabling nagle will cause small requests to slow down where as this approach adjusts. !batching_benchmark.png! > Coalesce intra-cluster network messages > --------------------------------------- > > Key: CASSANDRA-8692 > URL: https://issues.apache.org/jira/browse/CASSANDRA-8692 > Project: Cassandra > Issue Type: Improvement > Components: Core > Reporter: Ariel Weisberg > Assignee: Ariel Weisberg > Attachments: batching-benchmark.png > > > While researching CASSANDRA-8457 we found that it is effective and can be > done without introducing additional latency at low concurrency/throughput. > The patch from that was used and found to be useful in a real life scenario > so I propose we implement this in 2.1 in addition to 3.0. > The change set is a single file and is small enough to be reviewable. -- This message was sent by Atlassian JIRA (v6.3.4#6332)