Monitor performance for MediaPackage, CDN, and MediaTailor integrations
AWS Elemental MediaTailor requires effective monitoring to maintain optimal performance of your AWS Elemental MediaPackage and content delivery network (CDN) integration. This topic provides guidance on key metrics to track, monitoring tools to use, and how to set up alerts for proactive issue detection.
Before setting up monitoring, ensure your basic integration is working correctly. If you haven't completed your basic content delivery network integration setup, start with Integrate MediaTailor with MediaPackage and CDN . If you need to troubleshoot issues identified through monitoring, see CDN integration troubleshooting..
Key performance metrics
Monitor these essential metrics to ensure optimal performance of your MediaPackage and CDN integration:
CDN performance metrics
For comprehensive CDN performance metrics including cache hit ratio targets, origin request volume monitoring, and response time benchmarks, see Performance benchmarks for CDN and MediaTailor integrations in the CDN optimization guide.
Key EMP-specific considerations for CDN metrics:
- EMP cache-control headers
-
What to verify: Ensure your CDN honors EMP's cache-control headers for optimal TTL behavior
Expected behavior: Different content types should have different cache durations based on EMP's headers
For detailed guidance on EMP cache optimization, see Optimize CDN caching for MediaTailor and MediaPackage content delivery.
- Query parameter impact
-
What to monitor: Track how EMP-specific query parameters affect cache efficiency
Optimization target: Ensure only necessary EMP query parameters are included in cache keys
- CDN response times
-
What to monitor: Track response times for different content types (manifests vs. segments).
Target values:
-
Cached content: less than 100ms
-
Origin requests: less than 500ms
-
MediaPackage performance metrics
- Error rates
-
What to monitor: Monitor HTTP error rates from both your CDN and MediaPackage endpoints. Pay particular attention to 4xx errors, which might indicate configuration issues.
Key error codes:
-
400 errors: Often related to manifest filtering issues
-
404 errors: Might indicate routing or configuration problems
-
504 errors: Timeout issues, especially with LL-HLS
-
- Request volume and patterns
-
What to monitor: Track request patterns to MediaPackage endpoints to identify usage trends and capacity needs.
Patterns to watch:
-
Peak usage times
-
Geographic distribution of requests
-
Content type distribution (live vs. on-demand)
-
Latency metrics
- End-to-end latency
-
What to monitor: For LL-HLS implementations, monitor end-to-end latency from content ingest to viewer playback. High latency might indicate CDN configuration issues.
Target values:
-
LL-HLS: less than 3 seconds glass-to-glass latency
-
Regular HLS: less than 30 seconds
-
- Manifest generation time
-
What to monitor: Time taken by MediaPackage to generate manifests, especially with filtering applied.
Target values: less than 200ms for manifest generation
Monitoring tools and setup
Setting up comprehensive monitoring tools is essential for maintaining optimal performance and quickly identifying issues before they impact viewers. Without proper monitoring, performance degradation, cache inefficiencies, or integration problems might go unnoticed until viewers experience poor playback quality. The right monitoring setup provides visibility into all aspects of your MediaPackage and CDN integration.
Use these AWS services and tools to monitor your MediaPackage and CDN integration:
Amazon CloudWatch
Amazon CloudWatch provides the foundation for monitoring your MediaPackage and CDN integration by collecting and storing metrics from both services. Proper CloudWatch configuration ensures you have the data needed to identify performance trends, troubleshoot issues, and optimize your integration. Without CloudWatch metrics, you'll lack visibility into system performance and might not detect issues until they become critical.
Set up CloudWatch monitoring for comprehensive metrics collection:
-
MediaPackage metrics: Enable CloudWatch metrics for your MediaPackage endpoints to track request volumes, error rates, and response times.
-
CDN metrics: Configure CloudWatch to collect CloudFront metrics including cache hit ratios, origin request counts, and error rates.
-
Custom metrics: Create custom metrics for business-specific KPIs like viewer engagement or content popularity.
CloudWatch dashboards
Create comprehensive dashboards to visualize your metrics:
-
Overview dashboard: High-level metrics showing overall system health
-
CDN performance dashboard: Detailed CDN metrics including cache performance and geographic distribution
-
MediaPackage performance dashboard: MediaPackage-specific metrics including request patterns and error rates
-
Latency dashboard: End-to-end latency metrics for different content types and regions
Log analysis
Set up log analysis for detailed troubleshooting:
-
CDN access logs: Enable and analyze CDN access logs to understand request patterns and identify issues
-
MediaPackage CloudWatch logs: Monitor MediaPackage logs for errors and performance issues
-
Log aggregation: Use Amazon CloudWatch Logs Insights or third-party tools to analyze log patterns
Set up alerts and notifications
Alert configuration is crucial for proactive issue detection and resolution. Without proper alerts, issues might go unnoticed until they significantly impact viewer experience or cause service disruptions. Well-configured alerts help you identify and address problems before they affect your viewers, and ensure your team is notified of critical issues that require immediate attention.
Configure proactive alerts to identify issues before they impact viewers:
Critical alerts
Set up immediate alerts for critical issues:
-
High error rates: Alert when 4xx or 5xx error rates exceed 5% over a 5-minute period
-
Cache hit ratio drops: Alert when cache hit ratio falls below 70% for manifests or 85% for segments
-
High latency: Alert when end-to-end latency exceeds target thresholds
-
Origin request spikes: Alert when origin requests increase by more than 50% compared to baseline
Warning alerts
Set up warning alerts for trends that might indicate developing issues:
-
Gradual performance degradation: Alert when response times increase by 20% over a 30-minute period
-
Cache efficiency trends: Alert when cache hit ratios show declining trends over time
-
Unusual traffic patterns: Alert for unexpected changes in request volume or geographic distribution
Use monitoring data for optimization
Leverage monitoring data to continuously improve performance:
Regular performance reviews
-
Weekly reviews: Analyze weekly performance trends and identify optimization opportunities
-
Monthly capacity planning: Use traffic patterns to plan for capacity needs and CDN optimization
-
Quarterly architecture reviews: Evaluate overall architecture efficiency and identify areas for improvement
Common optimization actions
Based on monitoring data, consider these optimization actions:
-
Cache policy adjustments: Modify TTL values based on actual content update patterns. For detailed TTL optimization guidance, see Caching optimization for CDN and MediaTailor integrations.
-
Geographic optimization: Add CDN edge locations in regions with high traffic
-
Query parameter optimization: Remove unnecessary query parameters that fragment cache
-
Origin shield configuration: Implement origin shield in regions with high origin request volumes
For detailed monitoring guidance specific to MediaPackage, see Monitoring MediaPackage in the MediaPackage user guide.
