A single-page app serves static content from S3 through CloudFront and uses an EKS cluster for API requests. Users sometimes say the site is down even though the index page and EKS cluster checks look healthy. Which additional monitoring will proactively detect a site outage before users report it?

A CloudWatch Synthetics heartbeat monitor canary simulates a user accessing the website's FQDN. This proactively checks the entire user-facing stack, including CloudFront, S3, and the EKS backend, detecting issues like misconfigurations or partial outages that health checks might miss. If the canary fails to load the page, it indicates a site outage before users report it. Monitoring the EKS API endpoints with a Synthetics API canary would only check the backend, not the full user experience, so it wouldn't catch issues with the static content delivery or CloudFront. CloudWatch RUM (Real User Monitoring) collects data from actual user sessions, meaning it detects problems after users experience them, which is not proactive. RUM on EKS API endpoints would also only monitor the backend from the user's perspective, not the full site availability.