ECS Sporadic Workloads to Lambda

Opportunity Name:

ECS Sporadic Workloads to Lambda

AWS Resource Type:

Amazon ECS Fargate Service

Opportunity Description:

This Finder identifies Amazon ECS Fargate services with sporadic CPU activity that may be more cost-effective to run on AWS Lambda.

CloudFix analyzes CloudWatch CPU utilization metrics for continuously running ECS Fargate services and determines how frequently they are actively processing work. Services that show meaningful CPU activity in only a small percentage of 5-minute intervals are strong candidates for migration.

For workloads that are mostly idle, AWS Lambda’s pay-per-invocation pricing model can significantly reduce costs compared to running Fargate tasks continuously.

Criteria for identifying the opportunity:

The Finder identifies ECS Fargate services that meet the following criteria:

• The service runs on ECS Fargate
• The service is part of an ECS service
• The service is older than the configured lookback period
• The task definition CPU is within AWS Lambda limits
• The task definition memory is within AWS Lambda limits
• CloudWatch CPU utilization data is available for the lookback period
• The service is active in less than the configured percentage of 5-minute windows
• The estimated annual savings meet the CloudFix minimum savings threshold

CloudFix evaluates sporadic usage using the CloudWatch CPUUtilization (Maximum statistic) over 5-minute periods. Each interval is classified as active when CPU utilization exceeds a defined threshold. Services with low active time are flagged as potential Lambda migration candidates.

Potential savings (range in % on annual basis):

Savings depend on how frequently the service is active:

• Highly sporadic workloads: Can achieve 90%+ annual savings
• Moderately intermittent workloads: Can still deliver meaningful savings
• Near-continuous workloads: Typically not suitable for this optimization

Lambda eliminates the need to pay for idle compute, making it particularly effective for bursty or event-driven workloads.

What happens when the Fixer is executed?

This opportunity does not include an automatic Fixer.

CloudFix provides a detailed recommendation report that includes:

• Current ECS service CPU and memory configuration
• Active time percentage over the lookback period
• Current annual ECS cost
• Estimated annual Lambda cost
• Estimated annual savings
• Lambda compatibility notes
• Migration prerequisites that must be validated manually

Customers are responsible for reviewing the recommendation and performing the migration.

Is it possible to rollback once CloudFix implements the fixer?

CloudFix does not implement this change automatically.

If you migrate the workload manually and need to revert, you can redeploy the application back to ECS Fargate using the original service configuration.

Can CloudFix implement the fix automatically once I accept the recommendation?

No. This is an opportunity with a manual fix.

Does this fix require downtime?

Potentially, depending on the migration strategy.

Migrating from ECS Fargate to AWS Lambda requires application refactoring and deployment changes. Any service impact depends on:

• The migration approach
• Event trigger configuration
• Traffic cutover strategy

A well-planned migration (e.g., blue/green or canary deployment) can minimize or avoid downtime.

Additional Resources:

• AWS Fargate vs Lambda Decision Guide