Thanks Marco. Caching, caching and caching :) Next days I'm going to setup a QGIS server from master to test the improvements. I will also test its behaviour behind a MapProxy instance with tiling and a complex labeled layer, to see the outputs with meta tiling and tile buffering. I'm gonna stress it!
giovanni 2013/6/22 Marco Hugentobler <[email protected]> > >It is important that you test QGIS master not QGIS 1.8. QGIS 1.8 was > definitely slower than UMN, but in master there are performance > improvements in the server part. Marco >knows the details. > > There have been a lot of performance improvements in QGIS core (e.g. the > new raster system, svg cache, more performant coordinate reprojection, CRS > and Coordinate transform cache, project file cache, ...). > > Nore that the benchmark especially measures the scalability of the WMS > server (many concurrent requests). For scalability, it is important that as > much is CPU-based and data comes from RAM rather than from disk. While a > desktop user e.g. does not notice a request to the sqlite DB to lookup a > CRS definition, this small disk access had a big impact on scalability. > Similarly for SVG files, project files, coordinate transformations. All > these things are cached now on first usage. > > Regards, > Marco > > > On 21.06.2013 13:10, Andreas Neumann wrote: > >> Hi Giovanni, >> >> It is important that you test QGIS master not QGIS 1.8. QGIS 1.8 was >> definitely slower than UMN, but in master there are performance >> improvements in the server part. Marco knows the details. >> >> Andreas >> >> On Fri, 21 Jun 2013 12:09:18 +0200, G. Allegri wrote: >> >>> Hi, >>> I was reading the graphs (I do not understand german, sorry) of the >>> slides from Sourcepole about their banchmarks [1]. >>> I was surprised to see how QGIS Server outperforms UMN Mapserver, >>> because my experience was different. I haven't done comparisons in the >>> last months. In the past performances didn't appear better then >>> Mapserver. >>> >>> Marco (and the ohters from Sourcepole), could you share the >>> configurations adopted for the benchmark? Hardware, http server >>> configuration, caching, tiling, etc.? Something is written inside the >>> PDF, but Google Translator isn't doing a great work :( >>> >>> Thanks a lot, >>> Giovanni >>> >>> >>> [1] http://sourcepole.ch/assets/**2013/6/17/fossgis_2013_** >>> performanceoptimierte_wms_**dienste.pdf<http://sourcepole.ch/assets/2013/6/17/fossgis_2013_performanceoptimierte_wms_dienste.pdf> >>> [1] >>> >> >> > > -- > Dr. Marco Hugentobler > Sourcepole - Linux & Open Source Solutions > Weberstrasse 5, CH-8004 Zürich, Switzerland > marco.hugentobler@sourcepole.**ch <[email protected]> > http://www.sourcepole.ch > Technical Advisor QGIS Project Steering Committee > > > ______________________________**_________________ > Qgis-user mailing list > [email protected] > http://lists.osgeo.org/**mailman/listinfo/qgis-user<http://lists.osgeo.org/mailman/listinfo/qgis-user> > -- Giovanni Allegri http://about.me/giovanniallegri blog: http://blog.spaziogis.it GEO+ geomatica in Italia http://bit.ly/GEOplus
_______________________________________________ Qgis-user mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/qgis-user
