AWS Compute Optimizer Cheat Sheet

AWS Compute Optimizer is a service that analyzes the configuration and utilization metrics of your AWS resources to recommend optimal configurations that can reduce costs and improve performance for your workloads. It uses machine learning to identify patterns in your resource usage and provide actionable recommendations.

Core Functionality

• Analyzes historical data: It analyzes historical utilization metrics from Amazon CloudWatch to generate recommendations.

• Cross-account and cross-region: You can view findings and recommendations across all your AWS accounts and Regions from a central dashboard.

• Reduces Costs: Identifies over-provisioned resources and recommends smaller, cheaper alternatives.

• Improves Performance: Detects under-provisioned resources that may be causing performance bottlenecks and recommends more suitable options.

Supported AWS Resources

Compute Optimizer provides recommendations for the following types of resources:

• Amazon EC2 instances: Recommends optimal instance types based on CPU, memory, network, and storage I/O usage.

• Amazon EC2 Auto Scaling groups: Recommends optimal instance types for the group's launch configuration or launch template.

• Amazon EBS volumes: Recommends changes to volume type, size, and IOPS to improve performance and reduce costs.

• AWS Lambda functions: Recommends optimal memory size configurations based on historical invocation data to improve performance and lower costs.

Key Concepts

Recommendation Preferences

These are optional features you can activate to enhance or augment the recommendations generated for your resources.

• Enhanced Infrastructure Metrics: This paid feature extends the look-back period for utilization metrics analysis from the default 14 days to three months, providing more accurate recommendations for workloads with infrequent peaks.

• Inferred Workload Type: This feature analyzes your running instances to identify the types of applications running (e.g., web server, database) and estimates the effort required to migrate them from x86-based instance types to Arm-based AWS Graviton instances.

• AWS Graviton-based Instance Recommendations: Provides recommendations for migrating to Graviton instances, showing the potential price and performance impact.

Delegated Administrator

• Within an AWS Organization, you can delegate a member account to act as the administrator for Compute Optimizer.

• This delegated administrator can view and manage recommendations, set preferences, and opt-in member accounts for the entire organization.

• There can only be one delegated administrator per organization.

Dashboard and Reporting

• The main dashboard provides a centralized view of all optimization findings for your resources across all enabled AWS Regions.

• You can filter and view recommendations based on finding types, such as "Under-provisioned" or "Over-provisioned".

• Recommendations can be exported to an Amazon S3 bucket for further analysis or sharing.

Pricing

• Standard Recommendations: AWS Compute Optimizer is available at no additional charge. It analyzes CloudWatch metrics that are enabled by default for your resources.

• Enhanced Infrastructure Metrics: This is an optional, paid feature. You are charged per resource, based on the number of hours the resource runs and is being analyzed with the extended three-month look-back period.