You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by dkundo <dk...@yahoo.com> on 2011/10/03 15:01:58 UTC
Faceted query performance problem when group.truncate set to true
Hi,
in my (test) setup I have 200K distinct documents with each document having
5 historical version of it (so in total there is 1M documents).
In order to to retrieve the latest (or historical) version of the documents
I'm using the grouping functionality:
id:[0 TO N]&group=true&group.field=objid&group.limit=1&group.sort=id
desc
In addition I need to provide faceted search:
facet=on&facet.field=feature&facet.field=tag&facet.field=folder
For faceted search to provide correct results I should add
*group.truncate=true* to my query. But when I do so the query time increases
significantly: from ~70ms without this option to ~1700ms with this option
set to true.
Am I doing something wrong?
Is there another way of doing faceted search combined with result grouping?
Regards,
Dmitry
--
View this message in context: http://lucene.472066.n3.nabble.com/Faceted-query-performance-problem-when-group-truncate-set-to-true-tp3389690p3389690.html
Sent from the Solr - User mailing list archive at Nabble.com.
Re: Faceted query performance problem when group.truncate set to true
Posted by Martijn v Groningen <ma...@gmail.com>.
Hi Dmitry,
What is the fieldtype of field objid? Grouping works much slower on
non string fields.
Post grouped faceting in general slows down your search time, b/c it
is an expensive operation to compute the grouped docset.
However a qtime that is 100 times slower is a lot. I have noticed that
if I use group.truncate=true that the search time is ~1.5 times
longer than without.
Grouping requires more memory to be utilized in general. What is your
configured maximum heapspace (-Xmx vm option)?
Martijn
On 3 October 2011 15:01, dkundo <dk...@yahoo.com> wrote:
> Hi,
>
> in my (test) setup I have 200K distinct documents with each document having
> 5 historical version of it (so in total there is 1M documents).
> In order to to retrieve the latest (or historical) version of the documents
> I'm using the grouping functionality:
>
> id:[0 TO N]&group=true&group.field=objid&group.limit=1&group.sort=id
> desc
>
> In addition I need to provide faceted search:
>
> facet=on&facet.field=feature&facet.field=tag&facet.field=folder
>
> For faceted search to provide correct results I should add
> *group.truncate=true* to my query. But when I do so the query time increases
> significantly: from ~70ms without this option to ~1700ms with this option
> set to true.
>
> Am I doing something wrong?
> Is there another way of doing faceted search combined with result grouping?
>
> Regards,
> Dmitry
>
>
> --
> View this message in context: http://lucene.472066.n3.nabble.com/Faceted-query-performance-problem-when-group-truncate-set-to-true-tp3389690p3389690.html
> Sent from the Solr - User mailing list archive at Nabble.com.
>
--
Met vriendelijke groet,
Martijn van Groningen