[ 
https://issues.apache.org/jira/browse/HAWQ-971?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Shivram Mani updated HAWQ-971:
------------------------------
    Description: 
Most of the data formats on HDFS don't support projecting specific columns. For 
such formats we can still avoid resolving every individual field in the record 
and instead only resolve fields that need to be projected and send NULLs for 
other fields.
This task specifically focusses on improving the performance of Hive PXF 
profile which positively impacts both Hive and HiveORC profile

We will be padding null objects of the appropriate type as part of the resolver 
for fields that are not required to be projected as per the SQL query. Only 
fields required by the query will be projected/resolved in the appropriate 
manner

  was:
Most of the data formats on HDFS don't support projecting specific columns. For 
such formats we can still avoid resolving every individual field in the record 
and instead only resolve fields that need to be projected and send NULLs for 
other fields.
This task specifically focusses on improving the performance of Hive PXF 
profile which positively impacts both Hive and HiveORC profile


> Update PXF Hive Resolver to return partial content
> --------------------------------------------------
>
>                 Key: HAWQ-971
>                 URL: https://issues.apache.org/jira/browse/HAWQ-971
>             Project: Apache HAWQ
>          Issue Type: Sub-task
>          Components: PXF
>            Reporter: Shivram Mani
>            Assignee: Shivram Mani
>             Fix For: backlog
>
>
> Most of the data formats on HDFS don't support projecting specific columns. 
> For such formats we can still avoid resolving every individual field in the 
> record and instead only resolve fields that need to be projected and send 
> NULLs for other fields.
> This task specifically focusses on improving the performance of Hive PXF 
> profile which positively impacts both Hive and HiveORC profile
> We will be padding null objects of the appropriate type as part of the 
> resolver for fields that are not required to be projected as per the SQL 
> query. Only fields required by the query will be projected/resolved in the 
> appropriate manner



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

Reply via email to