cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Robert Stupp (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (CASSANDRA-9738) Migrate key-cache to be fully off-heap
Date Wed, 29 Jul 2015 09:58:04 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-9738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14645767#comment-14645767
] 

Robert Stupp edited comment on CASSANDRA-9738 at 7/29/15 9:57 AM:
------------------------------------------------------------------

[This cstar test|http://cstar.datastax.com/tests/artifacts/8128ee5e-35bf-11e5-804a-42010af0688f/graph]
is probably interesting. It compares 2.2, trunk, trunk w/ ParNewCMS, 9738, 9738 w/ ParNewCMS.
The reason for the additional ParNewCMS configs is that it was the only reasonable change
that can explain the weird throughput graphs (and huge GC pauses) for trunk.
After some offline discussion yesterday with [~benedict] and [~tjake], comparing G1 w/ ParNewCMS
we concluded that G1 is "the cause" of the weird graphs + long GC pauses for trunk. G1 requires
more head room to perform better. And 9738 gives it more head room (by moving stuff from heap
to off-heap).


was (Author: snazy):
[This cstar test] is probably interesting. It compares 2.2, trunk, trunk w/ ParNewCMS, 9738,
9738 w/ ParNewCMS. The reason for the additional ParNewCMS configs is that it was the only
reasonable change that can explain the weird throughput graphs (and huge GC pauses) for trunk.
After some offline discussion yesterday with [~benedict] and [~tjake], comparing G1 w/ ParNewCMS
we concluded that G1 is "the cause" of the weird graphs + long GC pauses for trunk. G1 requires
more head room to perform better. And 9738 gives it more head room (by moving stuff from heap
to off-heap).

> 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
>            Assignee: 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)

Mime
View raw message