[Benchmarking] Presentation abstract and benchmark contents

[Paul Ramsey]

The submission form only had room for one name, so it's mine. When we
have our official presenters, I'll ask the conference to change it
over. Are you going to be Mr Mapserver, Jeff?

P

On Fri, Jun 5, 2009 at 7:09 AM, Jeff
McKenna<jmckenna at gatewaygeomatics.com> wrote:
> As Daniel mentioned I seemed to have missed this abstract discussion while
> in transit - Paul which names/presenters did you put down for this abstract?
>
> (I assumed the submission deadline was moved to June 8th)
>
> -jeff
>
>
>
>
> Paul Ramsey wrote:
>>
>> According to the web site, the deadline is June 1, so I submitted
>> right away just to be on the safe side. We can edit the abstract
>> later. I submitted the version I edited:
>>
>> This is the latest installment in an annual series of benchmarks and
>> presentations that pit open source and other web mapping servers
>> against one another in a suite of performance tests.  The presentation
>> seeks to find out the optimal software configurations
>> and to compare the performance in different real world use cases, such
>> as large data handling, OSM like rendering, large raster serving and
>> - Hide quoted text -
>> WFS scalability.  MapServer and Geoserver will be compared, as well as
>> different back-end data sources, such as PostGIS and Shapefiles.
>>
>> [Note to reviewers, other servers may also be added to the test,
>> pending participation by community members.]
>>
>> On Mon, Jun 1, 2009 at 8:06 AM, Andrea Aime <aaime at opengeo.org> wrote:
>>>
>>> Hi all,
>>> I believe the most urgent thing we need to do, as far as the
>>> presentation is concerned, is to file an abstract of it.
>>> The deadline is June 8, and if we don't have the presentation
>>> accepted, there is no much else to discuss, at least for FOSS4G.
>>>
>>> We need an abstract, and we need a set of presenters that will be
>>> there.
>>>
>>> For GeoServer, I would be the one. Who's going to put his name
>>> on the presentation for the MapServer side? What about MapGuide or
>>> DeeGree, do we have anyone? (which reminds me, we should invite
>>> from the other sides as well... maybe we should send a mail on
>>> osgeo-discuss).
>>>
>>> Abstract wise, I'm not a native english speaker, so someone else
>>> will certainly do better, but here is an attempt:
>>>
>>> ---------------------------------------------------------------------
>>>
>>> Title:
>>> Comparing the Performance of OGC web services
>>>
>>> Abstract:
>>> This is the third installment of a series of presentations that
>>> compared the performance of Web Map Service at
>>> given at FOSS4G 2007 and FOSS4G 2008.
>>> The presentation seeks to find out the optimal software configurations
>>> and to compare the performance in different real world use cases, such
>>> as large data handling, OSM like rendering, large raster serving and
>>> WFS scalability.
>>>
>>> ---------------------------------------------------------------------
>>>
>>> Of course the abstract somehow sets the contents of the presentation.
>>> What will we do this year? I would say, a bit of what was done
>>> last year so that we can compare with last year results, and something
>>> new to keep people entertained.
>>>
>>> I would suggest:
>>> - rendering the usual tiger_texas layer over shapefile and postgis.
>>>  Big chunky layer with very simple style. Maybe this year we should
>>>  give up doing non antialised output... or do you think it's still
>>>  relevant?
>>> - kill the states thematic map and go for OSM like rendering of a
>>>  city using OSM data, using shapes and external symbols as needed
>>> - raster wise, maybe again MrSid data, so that we can see MapServer
>>>  progress in that area, and a big(-ish) mosaic of GeoTiff tiles?
>>>  Maybe we can get some of that climatic change data in shape for
>>>  the latter.
>>> - wfs wise, maybe a test with an assortment of spatial and attribute
>>>  filters against a postgis database that has indexes (to see how
>>>  well they are used?)
>>>
>>> Anything else you might want to test out? I would leave tile
>>> cache out this year, as the comparison between direct and tile
>>> cached output has been made last year already, so I think there
>>> is one free slot for another benchmark/topic.
>>>
>>> Cheers
>>> Andrea
>>>
>
>
> _______________________________________________
> Benchmarking mailing list
> Benchmarking at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/benchmarking
>