A huge paradigm shift has been occurring with businesses moving towards managing applications rather than worrying about managing the components of the Kubernetes cluster itself, such as the initial number and scaling of nodes, or other operational requirements. Let’s take a look at why you would need a managed Kubernetes cluster in the first place.
Some advantages of using this approach include:
- Reduced operational overhead.
- Security patches and cluster maintenance are automated.
- Unnecessary capacity planning is eliminated.
- Allows you to focus solely on development.
The idea of a managed Kubernetes cluster is that we just select a location for deployment, and *poof* everything is provisioned and managed by the provider. A minimal cluster setup of a single node pool with 2-3 nodes, which you are not able to access, gives a somewhat serverless feeling, but is it really?
Two solutions to achieve this goal are Google Kubernetes Engine’s (GKE) Autopilot and Elastic Kubernetes Service’s (EKS) Fargate. Let’s take a look at them both in a few different categories.