Amazon ECS Introduces High-Resolution Metrics for Faster Service Auto Scaling

Amazon ECS has released support for high-resolution (20-second interval) metrics and metric publishing optimizations, enabling faster response to load changes. According to AWS benchmark tests, scale-out trigger time improved from 363 seconds to 86 seconds (a 76% reduction), and total scaling time including new task provisioning improved from 386 seconds to 109 seconds (a 72% reduction).

With this update, users can gain three key benefits:

• Improved performance and reliability: Applications respond faster to sudden traffic spikes, reducing user latency or errors.
• Cost optimization: With faster scale-out speed, there's no need to pre-provision extra capacity (padding) in preparation for traffic surges, reducing compute costs.
• Simplified configuration: Aggressive scaling behavior that previously required complex configurations like step-scaling can now be implemented using just a Target Tracking policy.

This feature is available across all ECS compute options, including AWS Fargate, ECS Managed Instances, and Amazon EC2. To use high-resolution metrics, you need to select the 20-second resolution in the monitoring settings when creating or updating a service, and unlike standard resolution (60 seconds), this incurs additional CloudWatch costs.