4

I have a polygon geodata in postgresql and I am publishing out to geoserver the 1.7 million records. How long is too long when processing?

EDIT: Forgot to mention that I successfully published a database spatial view subset of this same data.

Side question, What is postgresql really doing during this process???

@Devdatta Tengshe, I have a polygon dataset with about 1.7 million records and I am just trying to publish the data to geoserver. I published a spatial view from a small table and a join to the large polygon dataset. This layer seems fine in openlayers. I just published another spatial view from a different small table joined to the large polygon dataset and it published but will not finish loading in openlayers.

|improve this question
  • You need to provide more details for us to answer this. Firstly what kind of processing are you doing? What have you done so far? And what exactly are you trying to ask here? – Devdatta Tengshe Feb 1 '13 at 3:23
  • I am currently using up 43 percent of my physical memory. I have a too many apps running on my laptop. Perhaps I just need to beef up first. – Justin Feb 1 '13 at 6:06
  • How are you consuming it in OpenLayers? as WMS or WFS? do you see any errors in geoserver? Have you configured a spatial index over the data? – Devdatta Tengshe Feb 1 '13 at 6:18
  • Try to use the Layers Preview in Geoserver, and let the data to load. You could also take a look at Geoserver log. – Niculita Mihai Feb 1 '13 at 6:43
  • Would I build a spatial index on the spatial view or the large parcels table? The view runs ok in qgis. I left the data loading for around 20 minutes. – Justin Feb 1 '13 at 6:51
4

Several things to think about with this amount of data

1) Create a Spatial index in PostgreSQL/PostGIS, like this

CREATE INDEX table_geom_idx ON tablename USING GIST (geometry_column name);

2) What scales are you showing the data at, something like 1.7 million features probably needs to be shown at a set scale range. In GeoServer you can use a Min and Max zoom threshold, e.g

<MinScaleDenominator>16000</MinScaleDenominator>
<MaxScaleDenominator>32000</MaxScaleDenominator>

You will need to add this to your SLD, look at the GeoServer SLD cookbook for further examples

http://docs.geoserver.org/latest/en/user/styling/sld-cookbook/index.html

3) Can you generalise your data at other scales?

4) Tweak postgresql.conf to make sure it is tuned for performance. Best resource I found was this

http://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server

I easily publish 10s millions of vector features with PostGIS and GeoServer, but you need to use the above tweaks to get it working efficiently

|improve this answer
  • I'll implement these tips. The problem seems to be with Geoserver/hardware and not postgresql as the views and large dataset works well in qgis. Thanks and I'll report back. – Justin Feb 2 '13 at 21:59
0

Not sure if I have to accept this as an answer. It is rather noobish in a few ways. I deleted the layers that came pre-published with an install of OpenGeo Suite and in the same breath, went to GeoWebCache and enabled seeding on the one layer of interest and just mouse clicked through...

So I'm not sure which one it was. But my data is now starting to move...had to let the initial caches get built.

|improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.