[
https://issues.apache.org/jira/browse/CASSANDRA-9738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14720486#comment-14720486
]
Robert Stupp commented on CASSANDRA-9738:
-----------------------------------------
It's just brain dumping an idea: we could change RowIndexEntry and related
IndexInfo plus its comparators to be "off heap compatible" (means work on a
ByteBuffer but not change the serialization format or algorithm itself). Having
this would also save polluting the heap with RIE+II objects when reading it
from disk as it could just load one ByteBuffer and operate on that.
In other words: make {{RowIndexEntry}} a wrapper for a {{ByteBuffer}} without
the need to deserialize any object structure during reads.
CASSANDRA-9754 could further optimize this to handle really big partitions?
But having this problem with big RowIndexEntries (lots of IndexInfo objects) I
guess migrating KC off-heap is out of 3.0?
> 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: 3.0 beta 2
>
>
> 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)
