[GeoServer 2.1.3 (standard non-native JAI), Oracle (Locator) 11.1.2, CentOS 64-bit, Java 1.6]
I'm working on a project that has previously mapped up to 50,000 point features and a similar number of line features connecting them. Performance from GeoServer has been great (two thumbs up). The dataset was recently pushed to 130,000 points and performance nose-dived. I'm in the middle of investigations into resource usage, the potential for memory / disk swapping etc, but I was hoping to get a feel from other GeoServer users on whether this was normal - perhaps what I'm seeing is standard? In a single WMS 1.1.1 request I am rendering 130,000 point features. Locations are stored in Oracle in WGS84 lat/lon with a default r-tree spatial index. Points are styled with the default point SLD (red square, no external graphic) and no filter criteria are applied in SQL other that an SDO_RELATE on the bbox. No projection occurs other than getting the lat/longs onto a 2D image. From looking at Oracle traces and GeoServer-GeoTools logs this is what I think happens: - Oracle receives a very basic query from GeoServer. Returning a result takes around 2.5 seconds (there could be separate Oracle performance issues here that need addressing, but for now I want to focus on GeoServer). - GeoServer seems to ask for the results in batches of 200 (Oracle reports 600+ occurences of waiting for instructions from the client) and renders each feature as it comes back. - After the Oracle query completes GeoServer takes 5 seconds to request and draw 130,000 features with default red square. 5 seconds seems like a long time for 130,000 features with very basic styling. Does this indicate a problem? I am yet to really dig into resource usage but I think I saw that Java was consuming several GB of memory, which I think pushes it into swap space and I know this will slow things down. Is it reasonable for Java to use so much memory for such simple rendering? I am hoping to try native JAI soon but my gut tells me that although this might help there could be a more fundamental problem here. Any thoughts or previous experience with this type of problem would be **very** much appreciated. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/5-seconds-to-render-130-000-points-typical-GeoServer-WMS-performance-tp4995984.html Sent from the GeoServer - User mailing list archive at Nabble.com. ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Geoserver-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geoserver-users
