....
EvE moves players between servers by the use of jump gates. It was an early solution to the idea of having everyone in the same universe, and their implementation is still a technical standout. For CIG, server meshing is intended to provide that function, but we actually see the same bottleneck - how much work can a given server handle?
For reply to the final question: probably a lot IF the right configurations are done correctly.
AWS works of course on a virtual infrastructure, there are several ESX solution (Microsoft Hyper-V, VMWare v-Shpere just for nnaming the most common...) and even if the implementation could be different, both does have a "scaling system feature" that allow Virtual server to request additional computational resurces (aka RAM and CPU core) if needed...
I don't know the limit for Hyper-V solution and even less the Hypervisor AWS is based on, but for v-sphere the maximum configurable prarameter for a single VM are 6Tb of RAM, and 256 v-Core (a v-core is like a core of a CPU) so they can be massive....
Of course you need to have the phisical resurces available to configure that and that while I've no doubt that Amazon can provide that kind of power, CIG need to rpovide a big enough kind of money, so the problem is not if the server can manage a freaking amount of data, but if CIG is willing to pay for that amount of power.
Oh and to be clear: no server don't need to have those parameter configured all the time: there are automation tool that can re-configure the Vitural server paramter adding or removing resurces while if needed according to pre-configured rules and limit to allow to limit the amount of resources needed globally while assuring the full power when and where needed.
