[jira] [Commented] (PHOENIX-2090) Refine PhoenixTableScan.computeSelfCost() when scanRanges is available

["James Taylor (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/PHOENIX-2090?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14995891#comment-14995891 ] 

James Taylor commented on PHOENIX-2090:
---------------------------------------

PHOENIX-2143 will help with this, as it'd let you query the stats table given a start and stop row key to see how many guideposts (i.e. bytes) would be traversed.

> Refine PhoenixTableScan.computeSelfCost() when scanRanges is available
> ----------------------------------------------------------------------
>
>                 Key: PHOENIX-2090
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-2090
>             Project: Phoenix
>          Issue Type: Sub-task
>            Reporter: Maryann Xue
>              Labels: calcite
>   Original Estimate: 120h
>  Remaining Estimate: 120h
>
> We should compute a more accurate cost based on the "scanRanges" so that we can better choose among these different indices.
> For example, if we have more than one indices concerning different index keys, for example IDX1 is indexed on column a, and IDX2 is indexed on column b, and our query is like "select x, y, z where a between 'A1' and 'A2' and b between 'B3' and 'B4'.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)