As in past years, OpenGeo is participating in the annual WMS benchmarking exercise that OSGeo sponsors. The exercise is a great opportunity for development teams to learn about where their project needs to concentrate development effect to make things faster and more compliant in the next year. Our coordinator this year (as in the past couple years) is Andrea Aime, who is working hard on both making sure GeoServer is properly configured for the tests, and is providing JMeter expertise to the entire exercise. The results of the benchmarking will be presented at FOSS4G 2010 next month.

In previous years, the exercise has been just MapServer versus GeoServer, and both projects have learned a lot in the process. Last year ESRI joined the process, but unfortunately ran out of staff time before the results were complete. This year the participation has exploded! The final number of teams that complete the testing might be smaller, but as it stands now the participants include: MapServer, GeoServer, CadCorp GeognoSIS, Constellation SDI, ERDAS Apollo, Mapnik, Oracle MapViewer, and QGIS mapserver.

As the participants get their servers online and configured to use identical data and styles, it becomes possible to compare their results visually. To make the comparison a little more dynamic, I wrote a small multi-map page that uses OpenLayers to view all the servers at once, looking at the same area.

WMS Browser

It’s really fun to see so many different implementations chew up the same data and styles and spit out the same map. Standards work? Standards work!

4 thoughts on “WMS Benchmarking

  1. BTW, did you invite ESRI to join this year WMS Benchmaking? I hope you invited ESRI. Did the decline to participate? Reasons? 🙂

  2. ESRI was invited. I don’t know that they ever officially declined. Unofficially I heard that they just weren’t willing to devote the staff time to it. Anyone who has been involved (and last year they devoted a good deal of time but still were not ready in time for the testing deadline) knows it is an awful lot of work. There was no reason a 3rd party couldn’t have brought ArcGIS Server to the table, but that would require… an awful lot of work!

    Any organization that is dying to see cross-compared benchmarks for particular platforms should consider funding the teams they are interested in. Makes it a lot easier on the participants, who could be using this time for billable work.

Comments are closed.