Set up monitoring and scaling for CDN and MediaTailor integrations

Effective monitoring and scaling strategies are crucial for maintaining optimal performance and viewer experience with your AWS Elemental MediaTailor content delivery network (CDN) integration. Implement these approaches to ensure your CDN integration performs reliably at scale.

Implement these monitoring and scaling strategies that follow:

1. Configure monitoring for these key metrics. For guidance on appropriate target values, see Monitoring MediaTailor with Amazon CloudWatch and consult your CDN provider's best practices:

   • CDN cache hit ratios (establish baseline metrics and targets based on your content type and delivery patterns)

   • Origin request volumes (monitor patterns during normal operation to establish baselines for anomaly detection)

   • Error rates by error type (define thresholds based on your service level objectives and MediaTailor best practices)

   • Response times (set appropriate latency targets based on your viewer experience requirements and geographic distribution)

   For detailed implementation instructions, see Creating CloudWatch dashboards to visualize your MediaTailor and CDN metrics together.

2. Set up alerts for unexpected traffic patterns or performance degradation. Configure thresholds based on your baseline metrics and service level objectives. For guidance on setting up alerts, see Creating Amazon CloudWatch alarms. Consider monitoring:

   • Significant deviations from baseline cache hit ratios (typically alert when falling under 85-90%)

   • Sudden increases in origin request volume (alert on 30% or greater increase from baseline)

   • Error rate spikes exceeding your defined thresholds (typically 1-2% for 4xx errors, 0.5% for 5xx errors)

   • Response time degradation beyond acceptable levels (typically >500ms for manifests, >200ms for segments)

   For implementation examples, see CloudWatch concepts for creating effective monitoring dashboards and alerts.

3. Create scaling plans for predictable high-traffic events. Your plans should include these key elements:

   • Pre-event capacity increases (24-48 hours before event start)

   • Gradual viewer ramp-up schedules (typically 10-20% of expected audience per 5-minute interval)

   • Regional capacity distribution based on audience (allocate capacity proportionally to expected regional viewership)

   • Post-event scaling procedures (maintain peak capacity for 30-60 minutes after event conclusion)

   For implementation guidance on scaling for high-traffic events, see Setting up a resilient end-to-end live workflow on the AWS Media Blog.

4. Implement failover and redundancy measures for critical streams, including:

   • Multi-region CDN deployments (at least two regions for critical content)

   • Backup origin servers (configured with automated health checks every 30 seconds)

   • Automated failover triggers based on health checks (typically after 2-3 failed checks)

   • Recovery procedures for different failure scenarios (documented with specific response time targets)

   For detailed implementation steps, see Optimizing high availability with CloudFront origin failover.