Amazon ECS launches high-resolution metrics for faster service auto scaling

Introduction of High-Resolution Metrics

Amazon Elastic Container Service (ECS) has launched high-resolution metrics aimed at improving service auto scaling. This feature allows ECS to adjust task counts in response to changing workload demands more quickly and efficiently. The metrics operate on a 20-second resolution, compared to previous resolutions, allowing for faster reactions to load changes.

Improvements in Scaling Timeliness

In AWS benchmarking tests, the time to trigger scale-out decreased dramatically from 363 seconds to 86 seconds, an improvement of 76%. Furthermore, the total time needed to scale and provision new tasks fell from 386 seconds to 109 seconds, representing a reduction of 72%. This shift facilitates a more responsive and dynamic service for users experiencing variable traffic.

Benefits of the New Metrics

Three primary benefits arise from implementing high-resolution metrics with ECS auto scaling: 1) Enhanced performance and reliability from quicker scaling that minimizes latency or failures, 2) Reduced baseline task counts, leading to lower compute costs while still managing traffic spikes, and 3) Simplified configuration processes, replacing complex scaling setups with easier target tracking policies.

Implementation and Compatibility

To utilize the new high-resolution metrics, users need to enable them within their ECS service settings and configure a target tracking scaling policy that utilizes these metrics. This updated auto scaling solution is compatible across all ECS compute options: AWS Fargate, ECS Managed Instances, and Amazon EC2.