There’s a strong argument that the server architecture needed to be better at launch, but then the game sold more than an order of magnitude better than it was expected to, so no one would have noticed that it scaled badly had the player count been in line with their design and testing.
Ah yeah that’s a tricky one. I guess as developers we’d all like to be ambitious and plan for millions of users but that sort of hardware and architecture takes time and money that might not be realistically in the budget/scope.
I’ve also not really got insight as to who would have a say on that kind of hardware, whether that’s PMs or devs. Probably higher-ups, right?
I think for something like this, you’d rent cloud servers as you’d expect the number of concurrent users to change over time and ideally would be able to spin up more capacity when you need it without having to have those machines available all the time. You still need some kind of system that decides when to order more capacity with enough warning that it’s actually available (you can tell AWS you want a VM immediately, but it still takes a couple of minutes to transfer your data onto it and boot it up, which is longer than people want to sit in a loading screen) and decides which servers to assign to which users.
There’s a strong argument that the server architecture needed to be better at launch, but then the game sold more than an order of magnitude better than it was expected to, so no one would have noticed that it scaled badly had the player count been in line with their design and testing.
Ah yeah that’s a tricky one. I guess as developers we’d all like to be ambitious and plan for millions of users but that sort of hardware and architecture takes time and money that might not be realistically in the budget/scope.
I’ve also not really got insight as to who would have a say on that kind of hardware, whether that’s PMs or devs. Probably higher-ups, right?
I think for something like this, you’d rent cloud servers as you’d expect the number of concurrent users to change over time and ideally would be able to spin up more capacity when you need it without having to have those machines available all the time. You still need some kind of system that decides when to order more capacity with enough warning that it’s actually available (you can tell AWS you want a VM immediately, but it still takes a couple of minutes to transfer your data onto it and boot it up, which is longer than people want to sit in a loading screen) and decides which servers to assign to which users.