AWS Compute Optimizer
AWS Compute Optimizer analyses the configuration and utilisation metrics of your resources and tells you which are over-provisioned, which are under-provisioned, and which are idle. It reports recent utilisation history alongside projected utilisation for each recommendation, so a rightsizing decision can be made on the price-performance trade-off rather than on instinct.
You must opt in before it analyses anything. It supports standalone accounts, member accounts, and the management account of an organization — opting in at the management account gives a view across every member account.
What it covers
Section titled “What it covers”- Amazon EC2 instances
- EC2 Auto Scaling groups
- Amazon EBS volumes
- AWS Lambda functions
- Amazon ECS services on AWS Fargate
- Commercial software licences
- Amazon Aurora and Amazon RDS databases
- NAT gateways
- Amazon DynamoDB
- Amazon ElastiCache
- Amazon MemoryDB
- Amazon DocumentDB
- Amazon WorkSpaces
- Amazon SageMaker
The database, Fargate and NAT gateway recommendations are the ones most likely to carry real money, and they are the ones missing from older summaries of this service that stop at EC2, EBS and Lambda.
How the analysis works
Section titled “How the analysis works”By default Compute Optimizer looks at the last 14 days of CloudWatch metrics. Two preferences change that:
- Enhanced infrastructure metrics is a paid feature that extends the lookback to 93 days for selected resources. Worth turning on for anything with a monthly or quarterly demand cycle, where 14 days is not a representative sample.
- Rightsizing recommendation preferences let you set CPU and memory headroom, utilisation thresholds, lookback periods and instance-family preferences at organization, account or Region level — so recommendations reflect your own risk appetite rather than the default.
EC2 memory utilisation is not visible to CloudWatch by default. Compute Optimizer can ingest external memory metrics from observability products such as Datadog and Dynatrace, which materially improves EC2 recommendations for memory-bound workloads.
Using the output
Section titled “Using the output”Findings appear in the Compute Optimizer console, in the console of the originating service (the EC2 console, for example), and through the API. Recommendations can be exported to S3 to track them over time or to feed a rightsizing backlog — a single export beats screenshotting a dashboard once a quarter.
Compute Optimizer findings also appear in AWS Trusted Advisor’s cost optimization category once you have opted in.