I had seen this problem in the distant past, which lead me to stop using the APR, but just got back around to re-testing it. The current test environment is
Windows 2012 R2, 64 bit Java 8 Tomcat 8.0.36 Tomcat Native APR 1.2.x GeoServer 1.9.1 Apache HTTPD V2.4 (as a proxy, using AJP to redirect traffic to Tomcat) The problem is that the system performance when the Tomcat Native library is present is MUCH worse than if it is not. In this case I benchmarked, and the average tile request transaction without APR takes about 0.15 seconds, and with APR it takes about 1.15 seconds. With a user grabbing 40-60 tiles per render, this takes a map draw from about 5 seconds to about a minute. Has anyone else observed this behavior? Is there any workaround? FWIW - the host system CPU use with the APR is about 60% of that without - a nice benefit if it didn't otherwise slow things down. ------------------------------------------------------------------------------ Developer Access Program for Intel Xeon Phi Processors Access to Intel Xeon Phi processor-based developer platforms. With one year of Intel Parallel Studio XE. Training and support from Colfax. Order your platform today. http://sdm.link/xeonphi _______________________________________________ Geoserver-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geoserver-users
