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

Posted to issues@hbase.apache.org by "mikhail (JIRA)" <ji...@apache.org> on 2011/05/05 10:45:04 UTC

[jira] [Commented] (HBASE-3855) Performance degradation of memstore because reseek is linear

    [ https://issues.apache.org/jira/browse/HBASE-3855?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13029241#comment-13029241 ] 

mikhail commented on HBASE-3855:
--------------------------------

When writing the scanner code for the new HFile format I thought about getting rid of linear scans by adding an array of key/value pair offsets at the beginning of each data block, so that seek/reseek can be implemented as a binary search. If we do have cases when the seek operation has to scan thousands of kvs, then I will certainly implement the binary search seek in the HFile V2 format.


> Performance degradation of memstore because reseek is linear
> ------------------------------------------------------------
>
>                 Key: HBASE-3855
>                 URL: https://issues.apache.org/jira/browse/HBASE-3855
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: dhruba borthakur
>            Assignee: dhruba borthakur
>
> The scanner use reseek to find the next row (or next column) as part of a scan. The reseek code iterates over a Set to position itself at the right place. If there are many thousands of kvs that need to be skipped over, then the time-cost is very high. In this case, a seek would be far lesser in cost than a reseek.

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