lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "David Smiley (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SOLR-10038) Spatial Intersect Very Slow For Large Polygon and Large Index
Date Wed, 25 Jan 2017 21:20:27 GMT

    [ https://issues.apache.org/jira/browse/SOLR-10038?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15838607#comment-15838607
] 

David Smiley commented on SOLR-10038:
-------------------------------------

Wow; 3 minutes for 10k points.  Roughly how many vertices are on the polygon?

https://cwiki.apache.org/confluence/display/solr/Spatial+Search
There are some tricks to speed up polygonal search.  One is setting "autoIndex" on the field.
 (no re-index required).  Just set that the true and leave it.  Then, fiddle with either distErr
or distErrPct to get the precision you want and no more than you need.  Lastly, fiddle with
prefixGridScanLevel  -- set it to the grid level that the internal algorithm switches from
recursive decomposition to scanning.  If you have a 20 level prefixTree, I recall it defaults
to 4 off from the bottom, thus 16.

> Spatial Intersect Very Slow For Large Polygon and Large Index
> -------------------------------------------------------------
>
>                 Key: SOLR-10038
>                 URL: https://issues.apache.org/jira/browse/SOLR-10038
>             Project: Solr
>          Issue Type: Improvement
>      Security Level: Public(Default Security Level. Issues are Public) 
>          Components: spatial
>    Affects Versions: 6.4.0
>         Environment: Linux Ubuntu + Solr 6.4.0
>            Reporter: samur araujo
>              Labels: spatialsearch
>
> Hi all, I have indexed the entire geonames points (lat/long) with JTS enabled, and I
am trying return all points (geonameids) within a certain polygon (e.g. Netherlands country
polygon). This query takes 3 minutes to return only 10.000  points. I am using only solr intersect.
no facets. no extra fitering.
> Is there any configuration that could slow down such a query to less than 300 ms?



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

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message