containers-roadmap [ECS] [request]: Predictive Auto Scaling for ECS Services

Community Note

Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
If you are interested in working on this issue or have submitted a pull request, please leave a comment

Tell us about your request We currently have a number of services deployed to ECS with EC2 Launch Type. Our traffic is cyclic and predictable (peak in day, slack at night, busier on certain days). We also run a number of batch jobs at night at specific times, which creates a predictable spike in load. We would like to use predictive auto-scaling to scale our ECS tasks to handle the anticipated workloads proactively.

Which service(s) is this request for? ECS

Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard? Due to the periodic and predictable nature of our workload, we would like our auto-scaling strategies to be tailored to our actual usage. EC2 Predictive Auto-Scaling seems to be the solution we need, but this is not available for ECS services.

ECS Service Auto-Scaling uses Application Auto-Scaling and only supports these policies:

Target Tracking Scaling
Scheduled Scaling
Step Scaling

The impact of not having predictive auto-scaling is slower (reactive) auto-scaling to sub-optimal capacity. This results in increased operational risk and higher operational costs.

Are you currently working around this issue? We're currently using ECS Target Tracking Scaling, but we would prefer to leverage AWS' predictive auto-scaling feature.

Additional context Stack Overflow thread: https://stackoverflow.com/questions/70024016/predictive-auto-scaling-for-aws-ecs-services

Attachments N/A