[jira] [Comment Edited] (CASSANDRA-9738) Migrate key-cache to be fully off-heap

["Robert Stupp (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-9738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14618972#comment-14618972 ] 

Robert Stupp edited comment on CASSANDRA-9738 at 7/9/15 6:28 AM:
-----------------------------------------------------------------

I did some [coding|https://github.com/snazy/cassandra/tree/9738-key-cache-ohc] just to see what this might buy us.
[cstar bench|http://cstar.datastax.com/graph?stats=e0c7514c-258f-11e5-8c01-42010af0688f&metric=op_rate&operation=3_read&smoothing=1&show_aggregates=true&xmin=0&xmax=138.16&ymin=0&ymax=185288.4] gives these nice numbers:

| || reads/sec || writes/sec || total gc time, 2nd read
|| trunk | 120k | 122k | 23sec (104 gcs, 222ms avg)
|| with CASSANDRA-9718 | 128k | 131k | 24sec (106 gcs, 222ms avg)
|| with key-cache off-heap | 163k | 136k | 2sec (111 gcs, 18ms avg)

EDIT: corrected accidentally misplaced GC info


was (Author: snazy):
I did some [coding|https://github.com/snazy/cassandra/tree/9738-key-cache-ohc] just to see what this might buy us.
[cstar bench|http://cstar.datastax.com/graph?stats=e0c7514c-258f-11e5-8c01-42010af0688f&metric=op_rate&operation=3_read&smoothing=1&show_aggregates=true&xmin=0&xmax=138.16&ymin=0&ymax=185288.4] gives these nice numbers:

| || reads/sec || writes/sec || total gc time, 2nd read
|| trunk | 120k | 122k | 2sec (111 gcs, 18ms avg)
|| with CASSANDRA-9718 | 128k | 131k | 24sec (106 gcs, 222ms avg)
|| with key-cache off-heap | 163k | 136k | 23sec (104 gcs, 222ms avg)


> Migrate key-cache to be fully off-heap
> --------------------------------------
>
>                 Key: CASSANDRA-9738
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-9738
>             Project: Cassandra
>          Issue Type: Sub-task
>            Reporter: Robert Stupp
>             Fix For: 3.x
>
>
> Key cache still uses a concurrent map on-heap. This could go to off-heap and feels doable now after CASSANDRA-8099.
> Evaluation should be done in advance based on a POC to prove that pure off-heap counter cache buys a performance and/or gc-pressure improvement.
> In theory, elimination of on-heap management of the map should buy us some benefit.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)