Well, one option is to seed the cache.  You can tell GWC to request
every tile up front so it will already be in the cache when a user
arrives.  Obviously this can use a lot of storage space for the cache,
and it imposes a heavy load on the back end while doing the seeding.
 
In terms of diagnosing the slow down.  You might want to look at the DB
logs compared to the requests to get an idea of when the slow down is
occurring (before or after GeoServer gets the data).  Styles with
rendering transformations might slow things down. If you are using
JDBCConfig to share configuration between the GeoServer nodes you might
end up with a lot of traffic to the database, particularly if you are
dealing with a lot of layers at once.
 
-- 
Kevin Michael Smith
smit...@draconic.ca
 
 
On Thu, Mar 17, 2016, at 09:53 AM, Jason King wrote:
> All,
>
> We have a setup which includes GWC that sits in front of 4 Geoserver
> instances.  We know the GWC caching is working because after an
> initial request, subsequent requests complete quickly from the cache
> until the tiles requested are cleared.
 Once the request goes to Geoserver again the requests are slow.  These
 requests are taking 15-20 seconds when hitting Geoserver.  This length
 of time is unacceptable.
>
> I’ve used Geoserver logging to test the queries being executed
> directly against the database and they’re completing in < 40ms so I
> know the problem isn’t there.  We’ve started running tests by
> requesting tiles directly from the Geoserver
 instance and we’re able to reproduce the long load times.  We are using
 a Default metatile size @ 4 x 4.  We have 2 Default Cached Gridsets for
 the 4236 projection and the 900913 projection, each at the 256x256
 dimension.
>
> Are there any tips and tricks for making this loading move faster?
> Any ideas on what could be causing the long load times?  I can provide
> additional details of our setup as well.
>
> Thanks in advance!
>
> *Jason King*
> *Software Engineer III*
> O. +1 757 546 3401
> M. +1 757 272 6938
> jason.king@q-free.com__
> www.q-free.com/openroads
> *Follow us:
*cid:image002.png@01D0BAF8.99C05980*  *cid:image004.png@01D0BAF8.99C05980*  
*cid:image006.png@01D0BAF8.99C05980*  *
> cid:image008.png@01D0BAF8.99C05980
> *Q-Free Open Roads*
> 103 Watson Road
>
> Chesapeake VA 23320
> cid:image008.png@01D0BAF8.99C05980
> cid:image010.jpg@01D0BAF8.99C05980
>
> ----------------------------------------------------------------------
> --------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785231&iu=/4140
> _________________________________________________
> Geoserver-users mailing list
> Geoserver-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/geoserver-users
> Email had 5 attachments:


>  * image001.png  1k (image/png)
>  * image002.png  1k (image/png)
>  * image003.png  1k (image/png)
>  * image004.png  1k (image/png)
>  * image005.jpg  7k (image/jpeg)
 
------------------------------------------------------------------------------
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
http://pubads.g.doubleclick.net/gampad/clk?id=278785231&iu=/4140
_______________________________________________
Geoserver-users mailing list
Geoserver-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-users

Reply via email to