[ https://issues.apache.org/jira/browse/HBASE-14613?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15207933#comment-15207933 ]
Anoop Sam John commented on HBASE-14613: ---------------------------------------- When working with HBASE-15180, discussed abt the check to know whether the MSLAB is enabled or not. In fact I made an initial patch with this check.. Now the Cells in the requests are decoded with sharing the same byte[] where the RPC reads the requests. This is with assumption that any way the cells will be again copied to MSLAB area just before adding it to Memstore. So when MSLAB is not enabled, we should continue with old way of decoding which allocated new byte[] just for one Cells content. With CMS, with out using MSLAB was not good.. Saw this in tests. I believe with G1GC, [~jingcheng...@intel.com] did some tests with MSLAB disabled, MSLAB on but no pool and MSLAB on with pool. Mind sharing those results? > Remove MemStoreChunkPool? > ------------------------- > > Key: HBASE-14613 > URL: https://issues.apache.org/jira/browse/HBASE-14613 > Project: HBase > Issue Type: Brainstorming > Reporter: Lars Hofhansl > Priority: Minor > Attachments: 14613-0.98.txt, gc.png, writes.png > > > I just stumbled across MemStoreChunkPool. The idea behind is to reuse chunks > of allocations rather than letting the GC handle this. > Now, it's off by default, and it seems to me to be of dubious value. I'd > recommend just removing it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)