I have just started learning about kubernetes and its concepts, so sorry if this is a lame question, however, I cannot seem to find the answer anywhere.

I understand that a node is a VM or a physical machine where the processes actually take place. Correct me if I am wrong, but what I have learnt is that kubernetes itself does not manipulate the number of nodes assigned to it. Therefore, I guess, if I want to scale up my application, I would need to add additional nodes to the cluster manually or by using an external cluster autoscaler.

So, if I have a cluster with two nodes (1 CPU each) all of the containers processes will take place on those two CPUs. If I have 4 replica sets of my pots, they all will be running on those 2 CPUs (2 replicas on each of the CPUs I guess).

Will there be any difference if I have 2, or 4, or 8 replica sets other than providing better roll-outs and more fault tolerance?

In general, is there any sense of having more replica sets than CPUs on the nodes?

And last, if a cluster autoscaler is enabled, would the number of replica sets effect the number of nodes?


That's a fantastic question, I'm still learning kubernetes myself so I can't give you any definite answers but in my experience I've found that you're pretty much spot on when it comes to replica sets. More replica sets pretty much just means more fault tolerance, assuming of course your nodes have the resources to support the extra load. In terms of an autoscaler, it will only take effect in the event that you have exceeded your available resource pool, usually in terms of available memory at least that's what's bottlenecked me in the past. Again I reiterate I'm not an architect and am also learning but thought I would share my experience.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.