Thank you very much Andrea, Simone and Balazs. Both good information and tips. I did not want to give to much details to start. But to answer your questions:
1. Yes I suspected because of the nature of the GetCapabilities that mosaics are the only way to go. 2. Mosaics also make sense since 90% of the data is orthoimagery, 5% DEM and about 5% a mix of vector and other formats (there are some "themes" that can break it into different groups, but it is meant to be all the latest imagery for whatever location is available). 3. The usage pattern is probably the key (see below). While there are a lot of files, we expect very low server load. A few users per day who want a specific set 5 to 10 images from a server. The hesitation in mosaics is two-fold. One was that I did not see any good automated way of creating the shapefiles for mosaics or ways to create the stores for them. I do admit to being a geoserver noobie, but I do read the documentation (600 pages) in detail, and what I am seeing is that we would have to create an automated way of creating a shapefile with outlines for each "theme" (10K or so files) and then submit this via the REST API to the create store (would like an example). The deeper reason for the hesitation in creating mosaics is that we are really not in the position to apply a blanket rule for the prioritization of overlapping images for a location. The user needs to be in control of this. That is the "advantage" to ArcGIS geoserver. filtering is dynamic on attribute type by the end user (cloud cover, quality, temporal age, or any other meta data), One thing that we are considering (thinking out the box) is to place an intercept proxy between geoserver and the GetMap requests. We would then create on the fly the store and layer and redirect the request to geoserver. This way would not prebuild the entire mosaic, just serve the items as needed (and do garbage collection after a several thousand requests). The trick will be creating the GetMap Requests for the clients (WorldWind, Google Earth, etc. ) and crafting a pseudo-getCapabilities that will hide what is going on behind the scenes. Any thoughts on this? -- View this message in context: http://osgeo-org.1560.n6.nabble.com/20-Million-Image-files-tp4503489p4508213.html Sent from the GeoServer - User mailing list archive at Nabble.com. ------------------------------------------------------------------------------ Virtualization & Cloud Management Using Capacity Planning Cloud computing makes use of virtualization - but cloud computing also focuses on allowing computing to be delivered as a service. http://www.accelacomm.com/jaw/sfnl/114/51521223/ _______________________________________________ Geoserver-users mailing list Geoserver-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geoserver-users
