Question

A SaaS company runs a customer-facing web application behind an Application Load Balancer with an Auto Scaling group of EC2 instances. Support tickets reveal that customers frequently detect broken checkout and login flows before the operations team does. Existing monitoring relies only on ALB target health checks (a TCP check on port 443) and CloudWatch CPU alarms. The team wants to detect user-facing functional failures and latency degradation for critical multi-step workflows proactively, ideally before real users are affected, with minimal application code changes. Which improvement best meets these requirements?

Accepted Answer

Create CloudWatch Synthetics canaries that script the checkout and login workflows on a schedule from multiple regions, and alarm on canary failures and latency thresholds.

Answer

Enable ALB access logs to S3 and build Athena queries with scheduled QuickSight dashboards to spot elevated 5xx error rates in checkout requests.

Answer

Change the ALB target group health check to an HTTP check against the /health endpoint and lower the healthy threshold to fail instances faster.

Answer

Enable Container Insights and increase the frequency of CPU and memory metric collection so scaling reacts sooner to load-related failures.