[ https://issues.apache.org/jira/browse/CASSANDRA-2252?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13002414#comment-13002414 ]
Stu Hood commented on CASSANDRA-2252: ------------------------------------- I'd like to go ahead and merge these patches together onto jbellis's version. Things that I will cherry pick from the alternate patch: * A generic Allocator interface, rather than referring to Memtables. There are various other places that we need to apply slabbing * The additional counter tests we added * Using the slab allocator's allocation count to determine throughput for memtables ---- The other realization we've had about slab allocation is that unless _all_ sources of fragmentation are eliminated, slabbing actually causes promotion failures to happen earlier, since it is harder to promote a slab into a fragmented oldgen. The other sources of fragmentation we suspect are: * IndexSummaries (easily slabbed) * the key and row caches (row cache tackled in CASSANDRA-1969) These can probably be defragged in separate tickets, as long as we commit to fixing them. > off-heap memtables > ------------------ > > Key: CASSANDRA-2252 > URL: https://issues.apache.org/jira/browse/CASSANDRA-2252 > Project: Cassandra > Issue Type: Improvement > Reporter: Jonathan Ellis > Assignee: Jonathan Ellis > Fix For: 0.8 > > Attachments: 0001-add-MemtableAllocator.txt, > 0002-add-off-heap-MemtableAllocator-support.txt, 2252-alternate-v2.tgz > > > The memtable design practically actively fights Java's GC design. Todd > Lipcon gave a good explanation over on HBASE-3455. -- This message is automatically generated by JIRA. - For more information on JIRA, see: http://www.atlassian.com/software/jira