Clarification on application level resource settings

On application level, you can set the amount of instances, RAM and CPU. This describes pods/replicas, right? So a process called “web” is being granted 1vCPU and GB RAM and can replicate itself between 1 and 4 times. The underlying instances in the AWS cluster (nodes) then must allow the maximum needed from this process, right?


Absolutely, you got it right. Here is a schema I draw to explain what will happen with your example:

Here is an example of 1 Kubernetes cluster with 3 nodes. The applications are dispatched according of the resources available across the cluster.