You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hbase.apache.org by "dhruba borthakur (JIRA)" <ji...@apache.org> on 2011/03/31 06:48:05 UTC

[jira] [Created] (HBASE-3718) Improve 'get' performance when row resides in memstore

Improve 'get' performance when row resides in memstore
------------------------------------------------------

                 Key: HBASE-3718
                 URL: https://issues.apache.org/jira/browse/HBASE-3718
             Project: HBase
          Issue Type: Improvement
          Components: regionserver
            Reporter: dhruba borthakur
            Assignee: dhruba borthakur


The regionserver uses a ConcurrentSkipList to store the KVs in the memstore. Although the order complexity of a lookup is O(n), still the latency to lookup a specific key in the memstore is very large, especially when the memstore is large and the KV.compare() method is costly.

One optimization is to investigate using a ConcurrentHashMap (instead of ConcurrentSkipList). The lookup and insertion cost is minimized. We can do it only for column-families that are marked as "do not support rangescans". 

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira