[
https://issues.apache.org/jira/browse/PHOENIX-992?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14005821#comment-14005821
]
Anoop Sam John commented on PHOENIX-992:
----------------------------------------
There is one difference though wrt performance (in using reverse scan vs
HTable.getRowOrBefore())
The HTable.getRowOrBefore() will do one parse of KVs within a HFile block.
The use of reverse scan(as in the patch) will do parse within an HFile block 2
times.
1st time for the seek to last row when the reverse scanner is opened.
The next() call will return the last KV but will a move to the previous KV.
This in effect will do a parse of KVs in the HFile block again.
Just telling the difference.
> Replace hTable.getRowOrBefore with ReverseScan to get maxKey in
> StatsManagerImpl
> --------------------------------------------------------------------------------
>
> Key: PHOENIX-992
> URL: https://issues.apache.org/jira/browse/PHOENIX-992
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.0.0, 5.0.0
> Reporter: Jeffrey Zhong
> Assignee: Jeffrey Zhong
> Priority: Critical
> Attachments: phoenix-992-v2.patch, phoenix-992.patch
>
>
> Below is the java doc of the function in HBase and we should remove it ASAP
> {noformat}
> * @deprecated As of version 0.92 this method is deprecated without
> * replacement.
> * getRowOrBefore is used internally to find entries in hbase:meta and makes
> * various assumptions about the table (which are true for hbase:meta but
> not
> * in general) to be efficient.
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
