Amazon ECS introduces new high-resolution metrics for faster service auto scaling

Container applications need to respond quickly to traffic spikes. If your ECS service waits minutes to scale up during a sudden surge, users experience slowdowns and timeouts. Amazon’s new high-resolution metrics feature addresses this timing challenge by allowing ECS auto scaling policies to react based on metrics collected at one-second intervals instead of the default one-minute intervals. This tighter feedback loop means your containers can scale up faster when demand increases and scale down more efficiently during quiet periods.

Here’s how it works technically: ECS auto scaling policies now support metrics pushed at higher frequency to CloudWatch, enabling the scaling engine to detect demand changes more rapidly. Your scaling policies can track real-time metrics like CPU utilization, memory usage, or custom application metrics (request count, queue depth, etc.) through three complementary approaches. Target tracking automatically scales to maintain your specified metric value—say, keeping CPU at 70%. Predictive scaling uses machine learning to forecast load based on historical patterns, scaling proactively before traffic actually arrives. Scheduled scaling handles known events by pre-scaling at predetermined times. With high-resolution metrics, each of these methods responds faster, reducing the lag between “traffic increased” and “new tasks are running.”

The practical benefit becomes clear in real-world scenarios. An e-commerce platform running a flash sale knows traffic will spike at 2 PM. Instead of getting hammered by requests and timing out while waiting for traditional one-minute metrics to trigger scaling, the service pre-scales with scheduled scaling, then uses high-resolution target tracking to fine-tune capacity minute-by-minute. A SaaS API experiencing unpredictable usage patterns no longer needs to over-provision resources “just in case”—predictive scaling ramps up based on learned patterns, while high-resolution metrics catch unexpected bursts within seconds. For teams managing multiple ECS services, this means lower costs from reduced over-provisioning and better customer experience from fewer delays.

If you’re currently managing ECS services, this feature is worth testing in non-critical environments first. Start by enabling CloudWatch high-resolution metrics on your container applications, then adjust existing scaling policies to use one-second metric granularity. Monitor the results carefully—more responsive scaling is powerful, but overly aggressive policies can create unnecessary churn. For teams already comfortable with CloudWatch and scaling policies, this is a straightforward way to tighten your auto scaling feedback loop without architectural changes.