Explanation
Why the waste happens and who it affects.
Fargate Spot provides the same performance characteristics at up to 70% lower cost, making it ideal for stateless jobs, batch processing, CI/CD runners, or retry-friendly microservices.
Billing model
The pricing dimensions that drive this cost.
Fargate is billed based on vCPU and memory usage per second. Standard Fargate pricing is significantly higher than Fargate Spot, which offers steep discounts for interruptible workloads.
How to detect
5 checks to find it in your estate.
Identify ECS workloads on Fargate that:
- Do not require continuous availability or persistent state
- Can tolerate retries or interruptions
- Are currently using standard pricing rather than Spot
- Run in non-production environments or are scheduled/batch in nature
- Have high "Fargate-vCPU-Hours" and "Fargate-GB-Hours" billed at on-demand rates
How to fix
3 ways to remove the waste.
- Add the FARGATE_SPOT capacity provider to the ECS cluster and set a capacity provider strategy on services or RunTask calls (Fargate Spot is ECS-only, not supported on EKS, and Linux-only)
- For mixed environments, use a capacity provider strategy with weights and a base across FARGATE and FARGATE_SPOT to route eligible tasks to Spot
- Monitor interruption rates and implement retry logic where necessary
Documentation
Vendor references for pricing and configuration.