[ 
https://issues.apache.org/jira/browse/SOLR-5478?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

David Smiley updated SOLR-5478:
-------------------------------
    Description: 
When the "fl" field list mentions a list of fields that all have docValues, and 
they are equivalent to the stored version (not true for multiValued field due 
to ordering), then we can fetch the values more efficiently from docValues than 
from stored values.  If this can't be done, might as well fetch them all from 
stored values when they have both (this is what happens today).

Former title: Speed-up distributed search with high rows param or deep paging 
by transforming docId's to uniqueKey via memory docValues

Originally the scope of this was just for the uniqueKey but now it's broader.
        Summary: Optimization: Fetch all "fl" values from docValues instead of 
stored values if possible/equivalent  (was: Speed-up distributed search with 
high rows param or deep paging by transforming docId's to uniqueKey via memory 
docValues)

> Optimization: Fetch all "fl" values from docValues instead of stored values 
> if possible/equivalent
> --------------------------------------------------------------------------------------------------
>
>                 Key: SOLR-5478
>                 URL: https://issues.apache.org/jira/browse/SOLR-5478
>             Project: Solr
>          Issue Type: Improvement
>          Components: Response Writers
>    Affects Versions: 4.5
>            Reporter: Manuel Lenormand
>             Fix For: 4.9, 6.0
>
>         Attachments: SOLR-5478.patch, SOLR-5478.patch, SOLR-5478 smiley 
> fl.fieldCacheFields.patch
>
>
> When the "fl" field list mentions a list of fields that all have docValues, 
> and they are equivalent to the stored version (not true for multiValued field 
> due to ordering), then we can fetch the values more efficiently from 
> docValues than from stored values.  If this can't be done, might as well 
> fetch them all from stored values when they have both (this is what happens 
> today).
> Former title: Speed-up distributed search with high rows param or deep paging 
> by transforming docId's to uniqueKey via memory docValues
> Originally the scope of this was just for the uniqueKey but now it's broader.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to