āāā Resource Right-Sizing
š” AWS Compute Optimizer
Compute Optimizer uses machine learning to analyze your resource utilization and recommends optimal AWS resource configurations ā helping you reduce costs and improve performance by right-sizing your infrastructure.
š Covers: Supported Resources Ā· How Recommendations Work Ā· Finding Wasteful ResourcesWhat is AWS Compute Optimizer?
AWS Compute Optimizer
A service that uses ML to analyze the utilization history of your AWS resources (CPU, memory, network) and recommends better-suited resource types. It identifies over-provisioned resources (paying for more than you need) and under-provisioned resources (too small, affecting performance). Think of it as an automatic right-sizing advisor.
š¦ Supported Resources
EC2 Instances
Recommends optimal instance type and size based on CPU, memory, network usage
Lambda Functions
Recommends optimal memory size based on execution patterns
EBS Volumes
Recommends optimal volume type (gp2āgp3) and IOPS provisioning
Auto Scaling Groups
Recommends optimal instance types for the group
ECS on Fargate
Recommends CPU and memory settings for containers
š What Compute Optimizer Tells You
| Finding Type | What It Means | Action |
|---|---|---|
| Over-provisioned | Resource is larger than needed. CPU/memory consistently <30% | Downsize to save money. Performance won't suffer. |
| Under-provisioned | Resource too small, causing throttling, timeouts, or crashes | Upsize to fix performance issues |
| Optimized | Resource is appropriately sized for its workload | No action needed |
| Insufficient data | Not enough history (needs at least 30 hours for EC2) | Wait for more data collection |
šÆ Exam Tip
Compute Optimizer is the answer for: "right-size EC2 instances", "find optimal Lambda memory without manual testing", "identify over-provisioned resources", "reduce costs by right-sizing". It's different from Lambda Power Tuning (which tests multiple configurations via Step Functions) ā Compute Optimizer passively analyzes existing usage patterns.