Which solution will meet these requirements with the LEAST operational overhead?
Use a self-managed node to supply compute capacity. Deploy the application to the new EKS cluster.
Use managed node groups to supply compute capacity. Deploy the application to the new EKS cluster.
Use AWS Fargate to supply compute capacity. Create a Fargate profile. Use the Fargate profile to deploy the application.
Use managed node groups with Karpenter to supply compute capacity. Deploy the application to the new EKS cluster.
Explanations:
Using a self-managed node requires manual management of the underlying infrastructure (e.g., EC2 instances), which increases operational overhead. The company does not want to manage the underlying compute infrastructure.
Managed node groups reduce some infrastructure management but still require attention to scaling, instance types, and capacity planning. This introduces more operational overhead compared to AWS Fargate, which abstracts infrastructure management entirely.
AWS Fargate provides serverless compute capacity where AWS manages the underlying infrastructure. This option reduces operational overhead as there is no need to manage compute resources, making it the most appropriate choice for the company’s requirements.
Managed node groups with Karpenter may help with scaling, but still requires managing compute nodes. It also adds complexity as Karpenter is a tool for provisioning and scaling, which increases operational overhead compared to using AWS Fargate.
From my point of view, the answer is:
Use AWS Fargate to supply compute capacity. Create a Fargate profile. Use the Fargate profile to deploy the application.