Skip to content
Refine

Glossary · Compute

What is Right-sizing?

Matching the instance type and size to actual workload demand — avoiding both over-provisioning (paying for idle capacity) and under-provisioning (performance pain).

By HabileLabs

Definition

Right-sizing

Right-sizing is the practice of choosing the smallest / cheapest instance type that still meets your workload's actual demand. The discipline applies to EC2 (most common), but also to RDS, ElastiCache, Redshift, OpenSearch, and any other AWS service with a sized-resource billing model.

AWS Compute Optimizer is the native tool that analyzes CloudWatch metrics over a 14-day window and recommends instance changes; Cost Explorer's rightsizing recommendations flag an EC2 instance as idle when maximum CPU stays below 1% for 14 consecutive days.

Why it matters for AWS cost

Over-provisioning is the single most common form of AWS waste, and the easiest to fix. Most teams provision for the worst-case load they imagined at design time, then forget to revisit when actual usage settles below that estimate.

The challenge isn't identifying the right size — Compute Optimizer and Refine both surface candidates. The challenge is the change-management process: which workloads are safe to resize? What's the rollback? Refine's recommendations come with the AWS API change in plain English and a confidence rating so you can prioritize safe wins first.

Common gotchas
  • !CPU is not the only metric — memory-bound workloads need memory metrics from the CloudWatch agent.
  • !Right-sizing across families (e.g. m5 → c5) may change network or storage characteristics. Test first.
  • !For burstable instances (t-family), credit balance matters more than instantaneous CPU.

Find waste like this in your own AWS account free

60-second IAM-role connection. Read-only. No signup needed for the bill analyzer.

Refine is built and supported by HabileLabs, an AWS Advanced Tier Services Partner.

Back to glossary15 terms in the AWS cost & FinOps glossary