4 ms·
The company I worked for back ended some of these datasets. Definitely a bunch of usgs imagery. At the time several racks of solaris servers were required to
by bnolsen 13y ago
The company I worked for back ended some of these datasets. Definitely a bunch of usgs imagery. At the time several racks of solaris servers were required to serve up the imagery using tomcat for the requests. Those machines were in canada. Windows was used for arcgis and oracle. Shortly after opteron was released and ibm came out with a 64bit jvm I moved the whole thing except for the esri portion over to a simple load balancer running apache and two mirrored dual opterons running fedora 64bit (2002). A total half rack which handled about a million image reqs a day or so. Each req involved decomposing, reading and stitching image mosaics together and sending out in under 10s. Unfortunately we were back ending more than just terraserver so I don't remember who had what specs. With all the memory alloc/dealloc going on inside the java process we were fortunate if it didn't require a full restart every 12 hours.
The funniest part of the whole thing was that I rewrote the engine from java over to a c++ stitcher and got request times under 1s on a dual p3 600. That was bypassing the esri server, oracle database and java stuff. At that time the umn mapserver started getting interesting as well although their image processing was glacial at the beginning. And then I was retasked and I left before being able to finish up and deploy.
- codeulike 13y agoWait, so you're saying that the Terraserver project that was supposed to show Windows and SQL Server capabilities was back-ended by a third party company using tomcat, oracle, java and linux. That doesn't make any sense. The OP-linked paper describes how it was all done with NT, IIS and SQL Server. http://msdn.microsoft.com/en-us/library/aa226316%28v=sql.70%29.aspx#msterra_hardware http://msdn.microsoft.com/en-us/library/aa226316%28v=sql.70%...