[
https://issues.apache.org/jira/browse/CASSANDRA-9738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14645807#comment-14645807
]
Benedict edited comment on CASSANDRA-9738 at 7/29/15 2:25 PM:
--------------------------------------------------------------
Looks like improved occupancy is also a factor (comparing run 2 to run 3). I'd
prefer to eliminate the malloc/free from cache maintenance before we hook OHC
into the key cache, but these numbers are pretty compelling and we should
probably consider it for 3.0.
I suspect CASSANDRA-8931 will permit us to deliver a more efficient and high
occupancy key-cache, but that shouldn't prevent us doing something in the
meantime.
was (Author: benedict):
Looks like improved occupancy is also a factor (comparing run 2 to run 3). I'd
prefer to eliminate the malloc/free from cache maintenance before we hook OHC
into the key cache, but these numbers are pretty compelling and we should
probably consider it for 3.0.
I suspect CASSANDRA-8930 will permit us to deliver a more efficient and high
occupancy key-cache, but that shouldn't prevent us doing something in the
meantime.
> 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)
