Originally Posted by
Zyrusticae
It's more than likely that the server hardware itself is inadequate to handle the sheer number of calculations being thrown at it at once.
This is not an easy problem to solve, mind you. For example, an often bandied-about solution is to increase the use of multithreading, but that requires that the code be programmed from the outset for parallelism - rewriting code to go from two, to three, to four threads is not at all a trivial task, and in some cases may simply be impossible (for example, in cases where data from one thread is dependent on receiving data from another, possibly causing stalling).
I have no idea how they handle this. It could be that each block is being handled by a single server shard, which would be a terrible idea and could be the cause of the heavy latency. But then, with 500 blocks out there, I can't say I'd blame them for taking that kind of route...
This is just speculation, mind you. I don't know the internal workings of Sega's server structure, I only know from experience with other, likely much smaller, server setups.
Connect With Us