Yep, this is why I stopped my thoughts on this whole thought process in my last email... This is exactly what I think the end goal of the WMS/WFS service tier should be... but first, let's worry about getting some _good_ multiresolution, well optimized datasets freely available... through process, so noone can turn off the spigot.
Baby steps. Although, if you want to define a grid for tiles, I'll follow your standards.... It's gotta start somewhere... Actually, if you wanted to be able to make use of the tile system through any application, all you would need to do would be to create a local proxy cascading web map server that would make the requests using the standardized tiles and then serve it up as a normal WMS to whatever client be it udig, mapserver, or even those people out in Redlands... ahh, it's so hard to stay on topic sometimes.... On 1/23/06, Chris Holmes <[EMAIL PROTECTED]> wrote: > Quoting David Bitner <[EMAIL PROTECTED]>: > > > There's already geotorrent with 33 datasets for full datasets. > > > > My thoughts for distributed map services would be to just devise a > > set > > grid at different resolutions and pass WMS requests through something > > like the coral cache. That way if other applications use the same > > WMS > > and make it so that their requests only use the same gridded tiles, > > you can start to develop a network of cached queries. Basically a > > fully server end version of what ka-map does for cacheing > I've been thinking about this issue a lot recently myself, and then > realized that Schuyler had already presented most of them: > http://mappinghacks.com/talks/osgeo_2005/distributed_wms_cache.ppt I > really like the idea of requesting gridded tiles, indeed I think it > could be very valuable to come up with a rough standard of how to > divide up the world, so that caches can start to build, across > applications. I think leveraging p2p networks is definitely the way to > go, and I'd like to see something integrated into WorldWind. They're > already rolling with multi-gig caches, that I believe are made from WMS > tile requests, but then they just sit on people's computers. Why not > query a p2p network first to see if anyone close by already has made > such a request. If not, then go to the server and get the dynamically > made response. > > I could also see one making a 'map accelerator', that just extends your > p2p client so that whenever your AJAXed out ka-map or mapbuilder makes > a WMS request it checks your local p2p server(/client, it's p2p so > they're the same) before actually sending the request out to the WMS > server. Once the global cache starts to build this has the great side > effect that one could put up a WMS server and be able to point all > sorts of users at it without fears of bandwidth and cost, since most > requests would just go to the p2p network. The reality is that putting > up a WMS/WFS server is just too expensive to keep fast if it's to get > popular at all. > > I have a friend working on a great new open source p2p client, who's > potentially very interested in abstracting out the functionality into a > nice api to support this. I believe it uses Jetty, so it's an actual > web server running on your computer, that links up to the p2p network. > So a variety of applications, AJAX like ka-map, openlayers and > mapbuilder, 3d like worldwind, and traditional gis like udig and gvsig, > could all program against the same api. And with it in place for ajax > apps you'd essentially get persistance in caching across sessions, as > the tiles would remain on your local machine. > > If this works I believe it could allow us wankers to get as much if not > more performance than google and all their data servers through out the > world, to serve up maps really fast. Perhaps we could even convince > them to make Google Earth a WMS tile p2p server/client, since I believe > one of their excuses for not supporting WMS out of box is that the > servers are unreliable. Combine this with a nuetral, global catalog > that anyone can update the metadata on wiki style, and also rate the > servers, then we're well on our way to a real geospatial web. > > Ok, must stop, made a new years resolution to drop all my longer > rambling postings in my blog instead of subjecting large lists of > people to them. > > best regards, > > Chris > > *** > Chris Holmes > The Open Planning Project > thoughts at: http://cholmes.wordpress.com > > ---------------------------------------------------------- > This mail sent through IMP: https://webmail.limegroup.com/ > _______________________________________________ Geowanking mailing list [email protected] http://lists.burri.to/mailman/listinfo/geowanking
