[
https://issues.apache.org/jira/browse/HBASE-19950?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16359542#comment-16359542
]
Anoop Sam John commented on HBASE-19950:
----------------------------------------
When we go with projecting only required column, only that CF related scan will
happen. We will consider that memstore only and HFiles of that store. Well
we have 'isFamilyEssential' been used also.. So first of all, the read will
try read from the required CF alone ie. the specified one as per this Filter.
Now when as per that it is include row, we will end up reading the other CFs
too.
bq.return compareValue(getCompareOperator(), getComparator(), c) ?
ReturnCode.SKIP : ReturnCode.INCLUDE;
Even we return INCLUDE_AND_SEEK_NEXT_ROW the read will happen... This is
what my concern over this..
You can try with testing abt the extra reads. I think that way it is going to
happen. Pls correct if wrong.
> Introduce a ColumnValueFilter
> -----------------------------
>
> Key: HBASE-19950
> URL: https://issues.apache.org/jira/browse/HBASE-19950
> Project: HBase
> Issue Type: Improvement
> Components: Filters
> Reporter: Reid Chan
> Assignee: Reid Chan
> Priority: Minor
> Attachments: HBASE-19950.master.001.patch,
> HBASE-19950.master.002.patch, HBASE-19950.master.003.patch,
> HBASE-19950.master.004.patch, HBASE-19950.master.005.patch,
> HBASE-19950.master.006.patch, HBASE-19950.master.007.patch
>
>
> Different from {{SingleColumnValueFilter}} which returns an entire row when
> specified condition is matched, this new filter will return the matched cell
> only. There're already some discussions in HBASE-19824.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
