exewu wrote:Nice post, good to know that we are close to a solution (or at least improvement) on the lag.
Well, yeah, that's the real question.
It is true that a good hosting site would lift the bandwidth limit, but it is worth noting that the current incarnation of the server won't be able to handle any amount of users for other reasons.
I'd guess that the server is currently capable of hosting somewhere between 400-600 online users CPU-wise, but depending on I/O bandwidth and the general performance of the routines for loading and unloading map data, that may or may not be limited to some lower number. That is one of the reasons why it hasn't been feeling enormously meaningful to try and find hosting. It may be possible to optimize the server further to drive that number up, however.
It should be mentioned in this context, however, that I recently managed to figure out a potentially workable method for parallelizing the server so that it can distribute its load over multiple CPUs, and that would also, to a very large extent, take care of the performance problems associated with map-loading. There would be quite some work involved in it, but assuming that I can make it work, I expect the server to be able to scale very well over multiple CPUs. A 4-way SMP system (e.g. a simple computer using a 4-core CPU) would then, I expect, be able to host somewhere between 1500-2000 users; and it should be able to continue scaling rather well with added CPUs.