[mapserver-users] Mapfile VS MapScript performance
Hi, I would to use MapServer to create WMS/WFS services. I think to compare 2 systems to produce services: 1) Apache + MapServer CGI + static mapfile (.map) 2) Apache + MapScript PHP + map configuration in cache Somebody has tested the performance between the mapfile and MapScript ? Thanks. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/Mapfile-VS-MapScript-performance-tp5014463.html Sent from the Mapserver - User mailing list archive at Nabble.com. ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] Mapfile VS MapScript performance
I have tested cgi against python mapscript (with mod_python) and the performance improvement, as expected, is massive. On Wed, Nov 7, 2012 at 8:53 AM, Spirifer ready...@gmail.com wrote: Hi, I would to use MapServer to create WMS/WFS services. I think to compare 2 systems to produce services: 1) Apache + MapServer CGI + static mapfile (.map) 2) Apache + MapScript PHP + map configuration in cache Somebody has tested the performance between the mapfile and MapScript ? Thanks. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/Mapfile-VS-MapScript-performance-tp5014463.html Sent from the Mapserver - User mailing list archive at Nabble.com. ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] Mapfile VS MapScript performance
As the cgi is robust and heavily tested I would think you would save massive amounts of developer time in not writing python wrapper code too! On Nov 7, 2012 8:42 AM, Umberto Nicoletti umberto.nicole...@gmail.com wrote: I have tested cgi against python mapscript (with mod_python) and the performance improvement, as expected, is massive. On Wed, Nov 7, 2012 at 8:53 AM, Spirifer ready...@gmail.com wrote: Hi, I would to use MapServer to create WMS/WFS services. I think to compare 2 systems to produce services: 1) Apache + MapServer CGI + static mapfile (.map) 2) Apache + MapScript PHP + map configuration in cache Somebody has tested the performance between the mapfile and MapScript ? Thanks. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/Mapfile-VS-MapScript-performance-tp5014463.html Sent from the Mapserver - User mailing list archive at Nabble.com. ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] Mapfile VS MapScript performance
The python wrapper is actually ~ 6 lines ;-) On Wed, Nov 7, 2012 at 9:47 AM, Mike Saunt mikesa...@gmail.com wrote: As the cgi is robust and heavily tested I would think you would save massive amounts of developer time in not writing python wrapper code too! On Nov 7, 2012 8:42 AM, Umberto Nicoletti umberto.nicole...@gmail.com wrote: I have tested cgi against python mapscript (with mod_python) and the performance improvement, as expected, is massive. On Wed, Nov 7, 2012 at 8:53 AM, Spirifer ready...@gmail.com wrote: Hi, I would to use MapServer to create WMS/WFS services. I think to compare 2 systems to produce services: 1) Apache + MapServer CGI + static mapfile (.map) 2) Apache + MapScript PHP + map configuration in cache Somebody has tested the performance between the mapfile and MapScript ? Thanks. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/Mapfile-VS-MapScript-performance-tp5014463.html Sent from the Mapserver - User mailing list archive at Nabble.com. ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] Mapfile VS MapScript performance
- testing against cgi isn't really an objective test, but even so the differences will iron out as soon as the actual workloads are not trivial (i.e. if you're just benchmarking getcapabilities then sure, mapscript keeping in memory your mapfile will be orders of magnitude faster than cgi. switching to complex map draws will show very similar performance between both options) - using fastcgi will somewhat iron out the differences on trivial workloads - certain mapscript scripts will leak non trivial amounts of memory, forcing server restarts periodically if you don't have an infinite amount of memory available. with all that said, my recommendation is to stick with cgi/fastcgi unless you have very specific needs that force you to use mapscript. -- thomas On Wed, Nov 7, 2012 at 9:42 AM, Umberto Nicoletti umberto.nicole...@gmail.com wrote: I have tested cgi against python mapscript (with mod_python) and the performance improvement, as expected, is massive. On Wed, Nov 7, 2012 at 8:53 AM, Spirifer ready...@gmail.com wrote: Hi, I would to use MapServer to create WMS/WFS services. I think to compare 2 systems to produce services: 1) Apache + MapServer CGI + static mapfile (.map) 2) Apache + MapScript PHP + map configuration in cache Somebody has tested the performance between the mapfile and MapScript ? Thanks. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/Mapfile-VS-MapScript-performance-tp5014463.html Sent from the Mapserver - User mailing list archive at Nabble.com. ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] Mapfile VS MapScript performance
On 11/07/2012 10:15 AM, thomas bonfort wrote: - certain mapscript scripts will leak non trivial amounts of memory, forcing server restarts periodically if you don't have an infinite amount of memory available. Hi Thomas, can you give more information about those memory leaks in MapScript? Jan ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] Mapfile VS MapScript performance
The performance point being made clear by Thomas I think I should mention a couple of (mostly in functionality) advantages that the WxS wrappers might have against cgi/fastcgi: 1. it is easy to extend the wrapper to add accounting, logging, caching. True, accounting and logging can also be accomplished simply by processing the web server log files but that requires extra software and might not be easy to impement in complex situations (think a cached request costs x and a non-cached y) 2. request preprocessing: filtering layers, redirecting 3. image/response prostprocessing: caching, assembling, watermarking,... Umberto On Wed, Nov 7, 2012 at 10:15 AM, thomas bonfort thomas.bonf...@gmail.comwrote: - testing against cgi isn't really an objective test, but even so the differences will iron out as soon as the actual workloads are not trivial (i.e. if you're just benchmarking getcapabilities then sure, mapscript keeping in memory your mapfile will be orders of magnitude faster than cgi. switching to complex map draws will show very similar performance between both options) - using fastcgi will somewhat iron out the differences on trivial workloads - certain mapscript scripts will leak non trivial amounts of memory, forcing server restarts periodically if you don't have an infinite amount of memory available. with all that said, my recommendation is to stick with cgi/fastcgi unless you have very specific needs that force you to use mapscript. -- thomas On Wed, Nov 7, 2012 at 9:42 AM, Umberto Nicoletti umberto.nicole...@gmail.com wrote: I have tested cgi against python mapscript (with mod_python) and the performance improvement, as expected, is massive. On Wed, Nov 7, 2012 at 8:53 AM, Spirifer ready...@gmail.com wrote: Hi, I would to use MapServer to create WMS/WFS services. I think to compare 2 systems to produce services: 1) Apache + MapServer CGI + static mapfile (.map) 2) Apache + MapScript PHP + map configuration in cache Somebody has tested the performance between the mapfile and MapScript ? Thanks. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/Mapfile-VS-MapScript-performance-tp5014463.html Sent from the Mapserver - User mailing list archive at Nabble.com. ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] Mapfile VS MapScript performance
Also using mapscript you can integrate mapserver with application by saving mapfile into database and letting application manage the mapserver configuration, grant access to authenticaded session clients You will lose speed but win maintenance. I do not really like to update the mapfile, then the tilecache then seed tiles again for each small change. 2012/11/7 Umberto Nicoletti umberto.nicole...@gmail.com The performance point being made clear by Thomas I think I should mention a couple of (mostly in functionality) advantages that the WxS wrappers might have against cgi/fastcgi: 1. it is easy to extend the wrapper to add accounting, logging, caching. True, accounting and logging can also be accomplished simply by processing the web server log files but that requires extra software and might not be easy to impement in complex situations (think a cached request costs x and a non-cached y) 2. request preprocessing: filtering layers, redirecting 3. image/response prostprocessing: caching, assembling, watermarking,... Umberto On Wed, Nov 7, 2012 at 10:15 AM, thomas bonfort thomas.bonf...@gmail.comwrote: - testing against cgi isn't really an objective test, but even so the differences will iron out as soon as the actual workloads are not trivial (i.e. if you're just benchmarking getcapabilities then sure, mapscript keeping in memory your mapfile will be orders of magnitude faster than cgi. switching to complex map draws will show very similar performance between both options) - using fastcgi will somewhat iron out the differences on trivial workloads - certain mapscript scripts will leak non trivial amounts of memory, forcing server restarts periodically if you don't have an infinite amount of memory available. with all that said, my recommendation is to stick with cgi/fastcgi unless you have very specific needs that force you to use mapscript. -- thomas On Wed, Nov 7, 2012 at 9:42 AM, Umberto Nicoletti umberto.nicole...@gmail.com wrote: I have tested cgi against python mapscript (with mod_python) and the performance improvement, as expected, is massive. On Wed, Nov 7, 2012 at 8:53 AM, Spirifer ready...@gmail.com wrote: Hi, I would to use MapServer to create WMS/WFS services. I think to compare 2 systems to produce services: 1) Apache + MapServer CGI + static mapfile (.map) 2) Apache + MapScript PHP + map configuration in cache Somebody has tested the performance between the mapfile and MapScript ? Thanks. -- View this message in context: http://osgeo-org.1560.n6.nabble.com/Mapfile-VS-MapScript-performance-tp5014463.html Sent from the Mapserver - User mailing list archive at Nabble.com. ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users