Benchmarking 2011/MeetingLog

Benchmarking 2011 IRC Meetings

• Wed February 23rd, 2011 [log: http://logs.qgis.org/foss4g/%23foss4g.2011-02-23.log]
  • Agenda:
    • Create plan for vector (OSM) and raster (DEM) data processing
    • Discuss testing methodology
    • final presentation discussion

• Wed February 16th, 2011 (log: http://logs.qgis.org/foss4g/%23foss4g.2011-02-16.log)
  • attendance:
    • Dane Springmeyer - Mapnik
    • Jeff McKenna - MapServer
    • Alex Wong - ERDAS
    • Pirmin Kalberer - QGIS mapserver
    • Mike Smith - Hardware / MapServer / Oracle MapViewer
  • Summary
    • final presentation discussion
      • springmeyer and jmckenna to submit abstract for presentation today or tomorrow
      • aaime sent message to jmckenna urging the use of the mailing list more by all teams
    • data sources discussion
      • springmeyer pointed out that some generalizing occurs with the osm2ogsql tool
      • springmeyer suggests 3 types of OpenStreetMap tests: data in shapefile (cloudmade extracts), data in PostGIS (osm2pgsql), and then a "best effort" (data in any format)
      • pirmin_k noted that osm2pgsql tool does not work well for QGIS
    • testing process discussion
      • covered in data discussion above

---

• Wed February 9th, 2011 (log: http://logs.qgis.org/foss4g/%23foss4g.2011-02-09.log)
  • attendance:
    • Marco Hugentobler - QGIS mapserver
    • Jeff McKenna - MapServer
    • Gabriel Roldan - GeoServer
    • Alex Wong - ERDAS
    • Mike Smith - Hardware / MapServer / Oracle MapViewer
    • Martin Daly - Cadcorp
  • Summary
    • data sources discussion
      • teams voted in support of using OpenStreetMap data for Colorado for vector
      • for raster, mpdaly suggests using DEMs
        • mdsmith suggests USGS Seamless NED or NAIP, and he also gets snowpack data daily
    • testing process discussion
      • teams support the plan to test only best effort, with data in format of your choice, but data must retain original detail (no generalizing)
      • continued discussion on testing tilecaching by servers...with the goal being to enhance how each server handles labelling on tiles

---

• Wed February 2nd, 2011 (log: http://logs.qgis.org/foss4g/%23foss4g.2011-02-02.log)
  • attendance:
    • Marco Hugentobler - QGIS mapserver
    • Mike Smith - Hardware / MapServer / Oracle MapViewer
    • Alex Wong - ERDAS
    • Daniel Morissette - MapServer
    • Martin Daly - Cadcorp
    • Jeff McKenna - MapServer
  • Summary
    • data sources discussion
      • unknown if suggested SPOT imagery can be distributed freely to the public
      • we should probably use local Denver/Colorado data
      • possibility of using U.S. Census TIGER data for Colorado
      • the benchmarking exercise in 2009 used TIGER data for Texas
      • mpdaly suggests using OpenStreetMap data
      • sample styling of MapServer using OSM data for Colorado already exists: map on http://www.mapserver.org/trunk/
    • testing process discussion
      • marco suggested that we have separate tests for point/line/poly, in addition to other tests
      • all teams voted to accept that we would only present a subset of the tests (we will decide what subset to use later)
      • mdsmith suggests testing tilecache seeding for each server...testing how the server's labeling engine handles edges of tiles

---

• Wed January 12th, 2011 (log: http://logs.qgis.org/foss4g/%23foss4g.2011-01-12.log)
  • attendance:
    • Andrea Aaime - GeoServer
    • Jeff McKenna - MapServer
    • Pirmin Kalberer - QGIS mapserver
    • Marco Hugentobler - QGIS mapserver
    • Mike Smith - Hardware / MapServer / Oracle MapViewer
    • Anne-Sophie Collignon - Erdas
    • Daniel Morissette - MapServer
    • Adrian Custer - Undetermined
    • Andreas Schmitz - Deegree
    • Martin Daly - Cadcorp
    • Jean-Francois Faudi - Data
    • Ivan Sanchez - Vocals
  • Summary
    • acuster: relayed that ESRI is still discussing internally about commitment
    • mdsmith: servers are now in our new datacenter and in racks. they are on ups + standby generator
    • discussion on data sources
      • jeffaudi: can supply 100 tiles of 1GB GeoTIFF imagery (SPOT)
    • all agreed data and styles need to be prepared soon
    • suggestion that testing methodology needs to be redesigned
    • teams will be allowed to join the exercise late, but must follow the rules of engagement (they can propose changes, but teams are not required to follow)