Gautam Parai created DRILL-6833:
-----------------------------------

             Summary: MapRDB queries with Run Time Filters with 
row_key/Secondary Index Should Support Pushdown
                 Key: DRILL-6833
                 URL: https://issues.apache.org/jira/browse/DRILL-6833
             Project: Apache Drill
          Issue Type: New Feature
    Affects Versions: 1.15.0
            Reporter: Gautam Parai
            Assignee: Gautam Parai
             Fix For: 1.15.0


Drill should push down all row key filters to maprDB for queries that only have 
WHERE conditions on row_keys. In the following example, the query only has 
where clause on row_keys:

select t.mscIdentities from dfs.root.`/user/mapr/MixTable` t where t.row_key=
(select max(convert_fromutf8(i.KeyA.ENTRY_KEY)) from 
dfs.root.`/user/mapr/TableIMSI` i where i.row_key='460021050005636')

row_keys can return at most 1 row. So the physical planning must leverage 
MapRDB row_key push down to execute the sub query, with its results execute the 
outer query. Currently only the inner query is pushed down. The outer query 
requires a table scan.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to