I have a set of about 250,000 polygons. Each of them contains a field
called geo, containing a geo_shape value (polygons). Querying against it
using a construct like this:
causes a query time of multiple seconds. I tried using both quad trees and
geohashes. I am querying for estates using polygons of streets, districts,
etc. Is there a way to speed up my query?
Assuming you are using 0.90, try to reduce the precision or tree_levels in
the mapping for the field to a few hundred meters.
If you aren't using 0.90, you should upgrade because that release contains
several geo_shape related fixes.
I have a similar data set (flickr neighborhood polygons) of about over 200K
shapes of varying sizes. Especially things like continents and other large
polygons are very hard on the index. I basically use several types with
different precision/tree_levels. For things like countries, I don't need a
lot of accuracy.
Jilles
On Sunday, May 26, 2013 4:33:01 PM UTC+2, Christian Sturm wrote:
Hello,
I have a set of about 250,000 polygons. Each of them contains a field
called geo, containing a geo_shape value (polygons). Querying against it
using a construct like this:
causes a query time of multiple seconds. I tried using both quad trees and
geohashes. I am querying for estates using polygons of streets, districts,
etc. Is there a way to speed up my query?
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.