Build here refers to software. "Server" term can refer to either hardware or software. This I am referring to software which is why the newest hardware is mentioned at the latter, they are two different things. So you have info on the coding and changes on the back end? Please do share. It just a logical assumption. How can amazon cloud architecture be using ancient hardware? What? You might want to read my posts again, you seem to be jumping all over the place. Huh, so you suggesting that anet did not pay them enough to scale appropriately?But you asking base on my reply, so I am not sure what I am getting wrong.This is way old, but my two cents. Scaling is only horizontal, meaning they can spin up more servers quickly for more instanced maps, ect. Cloud really shines here. But WvW maps can't be instanced the same and therefore can't scale the same way. Unless they bought dedicated VMs, there will be times when the WvW VM will be constrained just by the nature of being on shared infrastructure. I've seen a reduction in VM performance on Azure by 10 fold or more certain times of day. Getting 1/10 of the CPU and general performance in a layer beyond your control will result in unpredictable end user experiences, which is what I think we see often in GW2 WvW now. They should have kept WvW on prem on dedicated hardware IMHO.
