Maybe we just need a reasonable starting point and then work out the loads as we go? Would it seem logical when stress testing the server that you are only testing the server and not the network? If so, would it be reasonable to assume the client or test tool is run on the server itself and not as a remote client?
Many people do not have control over how much network bandwidth they need. We do however have much more control over our hardware configuration. Though I do pity the poor soul who puts together a huge quad core server but only has a 9600bps modem for their network connectivity! Regardless, if you are trying to stress test a server config to get a good idea of how many avatars it might be able to host for some given need like a dance party or training session would you first want to make sure the server can theoretically handle the load before worrying about the network bandwidth? Would a separate method for determining network per X many avatars be needed? If you build a server and know it can handle 35 avatars and you need Y amount of bandwidth per 10 avatars, can you at least predict the reliability of being able to accomplish some event? -Robert -- View this message in context: http://opensim-users.2152040.n2.nabble.com/Load-Testing-OpenSim-tp5197654p5226269.html Sent from the opensim-users mailing list archive at Nabble.com. _______________________________________________ Opensim-users mailing list [email protected] https://lists.berlios.de/mailman/listinfo/opensim-users
