[
https://issues.apache.org/jira/browse/HBASE-13370?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14480946#comment-14480946
]
ramkrishna.s.vasudevan commented on HBASE-13370:
------------------------------------------------
Just seeing this latest comment,
Your PE has 10 columns? So is that a modified version of the PE? Because by
default when we do sequentialWrite it writes data with one Column.
So am not sure what is your scan1000 doing. Is it again customised?
RandomScanWithRange10000Test by default adds a the default family and qualifier
explicitly thus every time leading to additional seek due to the explicit
column tracker. This change would help you in making that addColumn
configurable thus reducing one more seek.
> PE tool could give option for using Explicit Column Tracker which leads to
> seeks
> --------------------------------------------------------------------------------
>
> Key: HBASE-13370
> URL: https://issues.apache.org/jira/browse/HBASE-13370
> Project: HBase
> Issue Type: Improvement
> Affects Versions: 1.0.0, 1.0.1
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
> Attachments: HBASE-13370.patch, HBASE-13370_1.patch,
> HBASE-13370_1.patch
>
>
> Currently in PE tool all the scans and gets adds explicitly the columns to be
> scanned. The tool by default adds only one Qualifier. Doing this addColumns
> leads to Explicit Column Tracker which does seeks frequently. If we want to
> know a simple scan performance as a basic scenario then we should have the
> option to add this columns explicitly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
