Difference between kubectl apply and kubectl create

Question

From my understanding of the documentation: kubectl apply = kubectl create + kubectl replace. So, if I'm creating resources for the first time, I should use kubectl create. If I want to update something that already exists, I use kubectl replace. If I want to do both then I do kubectl apply.

I just want to know why are there three different commands doing essentially the same thing? Or is there some key difference that I'm missing?

Answer 1

These are two very different approaches. kubectl create uses Imperative Management. Here You specify what you want to create, delete or replace.

kubectl apply uses what we call Declarative approach. Here we tell the api How we want the cluster to look like. So even if you've applied changes to a live object, your changes are maintained.

More information on object management in kubernetes Kubernetes Object Management.