[ 
https://issues.apache.org/jira/browse/CALCITE-3442?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

yunfeng updated CALCITE-3442:
-----------------------------
    Description: 
 For elasticsearch-adapter,   when  ElasticsearchAggregate  involved , set 
_source = false and size = 0 , Elasticsearch `FetchPhase` would still be 
executed against the current  search process and visit the Lucene 
stored_fields, which would lead to performance declined dramatically.  
We can set `stored_fields = _none` prohibit this such behavior entirely, and 
skip the Elasticsearch's `fetch phase` which would bring time-consuming  `fetch 
phase` for  metadata fields such as '_id' etc. 

  was:
 For elasticsearch-adapter,   when  ElasticsearchAggregate  involved , set 
_source = false and size = 0 , Elasticsearch `FetchPhase` would still be 
executed against the current  search process and visit the Lucene 
stored_fields, which would lead to performance declined dramatically.  
We can set `stored_fields = _none` prohibit this such behavior completely


> set `stored_fields = _none_` prohibit `FetchPhase` get be involved and 
> improve the performance for Elasticsearch
> ----------------------------------------------------------------------------------------------------------------
>
>                 Key: CALCITE-3442
>                 URL: https://issues.apache.org/jira/browse/CALCITE-3442
>             Project: Calcite
>          Issue Type: Improvement
>          Components: elasticsearch-adapter
>            Reporter: yunfeng
>            Priority: Major
>              Labels: performance, pull-request-available
>             Fix For: next
>
>         Attachments: elasticsearch-adapter.diff
>
>          Time Spent: 40m
>  Remaining Estimate: 0h
>
>  For elasticsearch-adapter,   when  ElasticsearchAggregate  involved , set 
> _source = false and size = 0 , Elasticsearch `FetchPhase` would still be 
> executed against the current  search process and visit the Lucene 
> stored_fields, which would lead to performance declined dramatically.  
> We can set `stored_fields = _none` prohibit this such behavior entirely, and 
> skip the Elasticsearch's `fetch phase` which would bring time-consuming  
> `fetch phase` for  metadata fields such as '_id' etc. 



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to