AWS Compute Optimizer: The Recommendations Are Good. The Implementation Is on You.
23-05-2026 / Cost optimization strategies / 6 mins.
Here is a pattern that repeats across almost every AWS team: someone runs AWS Compute Optimizer, looks at the recommendations, thinks "yeah, those are probably right," and then does nothing for the next three months. The recommendations expire, new instances get added, and the cycle starts again.
It is not a discipline problem. It is a tool design problem. Compute Optimizer is excellent at what it does — and completely silent about everything it does not do.
What Compute Optimizer actually analyzes
Compute Optimizer ingests 14 days of CloudWatch utilization data for your EC2 instances, Lambda functions, EBS volumes, and ECS tasks. Based on real CPU, memory, network, and disk I/O patterns — not just CPU like older tools — it generates right-sizing recommendations with a projected performance risk rating.
For EC2, a typical recommendation looks like: Your m5.2xlarge is running at 8% average CPU utilization. An m5.large would handle your workload at Low performance risk and save you $X/month.
Those recommendations are genuinely good. The methodology is solid, the data is real, and the performance risk classification is usually honest.
What it does not cover
The bigger savings lever: commitments
Right-sizing is one piece of the AWS cost puzzle — and often not the largest one. For most companies with stable workloads, Reserved Instances and Savings Plans represent a bigger savings opportunity than right-sizing. AWS offers 20-40% discounts on EC2, RDS, and other services through these commitment-based models.
Compute Optimizer does not touch this layer at all. It optimizes how big your instances are — not how much you are paying per hour for them.
Idle and orphaned resources
Compute Optimizer analyzes running resources. It will not flag unattached EBS volumes, unused Elastic IPs, stopped instances that still have active RDS databases attached, or forgotten Load Balancers. All of these accumulate quietly on your bill.
Storage and data transfer
S3 storage class mismatches, cross-region data transfer costs, overprovisioned RDS storage — none of this is in scope for Compute Optimizer.
The implementation problem
Even when the recommendations are right, acting on them is harder than it looks. Resizing a production EC2 instance means validating the recommendation against actual application requirements, coordinating a maintenance window with the team, stopping the instance, changing the instance type, restarting, and monitoring for a week to catch any regression.
For a team managing 30-50 instances, that is days of engineering time. And it has to be redone every few months as workloads evolve. This is exactly why teams see the recommendations, agree with them, and still do not act.
Using Compute Optimizer the right way
The right frame for Compute Optimizer is: diagnostic tool, not complete solution. Run it to quantify your right-sizing opportunity, then pair it with:
- AWS Cost Explorer RI/SP recommendations — to understand the commitment-based savings opportunity, which is usually larger.
- AWS Trusted Advisor — for idle resource detection (requires Business or Enterprise support).
- A regular review cadence — recommendations go stale as new instances are added and usage patterns shift.
Where Frust fits in
While Compute Optimizer focuses on right-sizing, Frust focuses on the commitment layer — which for most companies is the higher-impact opportunity and requires zero operational changes to your workloads.
Frust connects to your AWS account read-only, identifies which Reserved Instances and Savings Plans make sense for your usage patterns, and purchases them on its own accounts. You get the discount on your bill without touching a single instance, without a maintenance window, and without taking on a multi-year commitment yourself.
The model: Frust charges 20% of the savings it generates. No savings, no charge.
Get a free savings estimate — connect your account and get a concrete projection within 48 hours.
See how much your company could save on AWS
Connect your account, get a savings estimate in less than 48 hours. No upfront commitment.

