It’s also dependent on what those 1000 are. Is it 1 app x 1000 instances or is it 500 apps x 2 instances. The former with 1 or 2 different apps, not much would

It’s also dependent on what those 1000 are. Is it 1 app x 1000 instances or is it 500 apps x 2 instances. The former with 1 or 2 different apps, not much would change, when you run lots of different apps though, then configuration of routes between different physical hosts and other odd network effects can start occurring. Rolling out updates takes significantly more time as you just have a larger number of nodes to get through, which means you’ll very likely have multiple different versions of your app running at once, potentially expecting different database schemas, migrations etc. So I guess I'm saying it’s mostly about the ops more than the underlying infrastructure, but its always both and there’s always something that’s a bottleneck somewhere