[Benchmarking] 2013 Shootout?

[thomas bonfort]

On 9 April 2013 14:54, Iván Sánchez Ortega <ivan at sanchezortega.es> wrote:

> On Martes, 9 de abril de 2013 14:35:14 Oliver Tonnhofer escribió:
> > Tile servers are mostly just serving static files and all implementations
> > should be able to push 1000 tile/s even on a small virtual server. This
> > should be fine for 99.9% of all setups (even OSM peaks below 5000
> tile/s).
> > Need more performance? Put a varnish in front of it. Done. Things you
> > can't do for WMS. So, that is IMO not really worth benchmarking.
>
>
> I just want to see the throughtput of Mapserver servind WMS compared to the
> throughput of Mapproxy serving WMS on top of Mapserver serving WMS. No
> serving
> tiles involved anywhere, and I don't care about the caching internals.
>
> And when I say "Mapserver WMS vs Mapproxy+Mapserver WMS" I also mean
> "Geoserver WMS vs GeoServer+Mapproxy WMS", "Cadcorp WMS vs Cadcorp+Mapproxy
> WMS", and so on.
>

Not a very informative test imho, as the result will be highly dependant on
the complexity of the underlying data. e.g:
- render an area with no data: the wms servers will be faster (they just
have to compress a png of the requested size, whereas a tile proxy will
have to do that *plus* decompress the individual png tiles that compose the
area)
- render an area with highly complex data (think whole of europe with an
osm2pgsql dump) and a tile proxy will be thousands of times faster.

--
thomas


>
>
> Best,
> --
> Iván Sánchez Ortega <ivan at sanchezortega.es> <ivan at geonerd.org>
> _______________________________________________
> Benchmarking mailing list
> Benchmarking at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/benchmarking
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/benchmarking/attachments/20130409/af7d0e9e/attachment.html>