Suboptimal Use of On-Demand Instances in a Non-Production EKS Cluster

Service Category

Compute

Cloud Provider

AWS

Service Name

AWS EKS

Inefficiency Type

Inefficient Architecture

Explanation

Running non-production clusters solely on On-Demand Instances results in unnecessarily high compute costs. Development, testing, and QA environments typically tolerate interruptions and do not require the continuous availability guaranteed by On-Demand capacity. Introducing Spot-backed node groups in non-production environments can significantly reduce infrastructure expenses without compromising business requirements.

Relevant Billing Model

EC2 instances backing an EKS cluster are billed based on instance-hour usage. On-Demand Instances incur premium rates compared to Spot Instances, which offer steep discounts in exchange for potential interruptions. Non-production workloads can often tolerate Spot interruptions, making them ideal candidates for cost optimization.

Detection

Identify non-production EKS clusters based on environment tagging, naming conventions, or cluster metadata
Review node group configurations to determine whether all nodes use On-Demand Instances
Assess workload criticality and tolerance for Spot interruptions based on application requirements
Analyze cluster autoscaler and provisioning policies to evaluate readiness for Spot adoption
Confirm plans with development and infrastructure teams before introducing Spot-backed capacity

Remediation

Provision new node groups, launch templates, or provisioners configured with Spot Instances. Migrate non-critical workloads to the Spot-backed nodes while retaining On-Demand capacity where appropriate for sensitive services. Monitor workload stability post-migration and tune autoscaling policies to optimize for cost and resilience. Update provisioning standards to prioritize Spot usage for future non-production clusters.

Relevant Documentation