To diagnose latency issues with AWS Global Accelerator, you should monitor several key metrics and use AWS monitoring tools to get detailed insights:
Key Metrics to Monitor
1. Round-Trip Time (RTT) / Latency
RTT measures the time it takes for a packet to travel from the client to the endpoint and back. It is the primary metric to assess network latency and is influenced by physical distance and network conditions. Monitoring RTT helps identify where latency is introduced in the path[1].
2. Throughput
This measures the amount of data or number of packets delivered over a time period. Low throughput can indicate network congestion or endpoint performance issues affecting latency[1].
3. Network Jitter
Jitter is the variability in latency over time. High jitter can cause inconsistent application performance and is often a sign of network instability or congestion[1].
4. Packet Loss
Packet loss occurs when packets fail to reach their destination. Even small amounts of packet loss can significantly increase latency due to retransmissions and degraded TCP performance[1].
5. Endpoint Health and Availability
Global Accelerator continuously monitors the health of endpoints using TCP, HTTP, and HTTPS health checks. Endpoint failures or degraded health status can cause traffic to be rerouted, impacting latency[7].
AWS Monitoring Tools and Logs
- Amazon CloudWatch Metrics and Alarms
Global Accelerator automatically reports metrics to CloudWatch once traffic flows through the accelerator. Key CloudWatch metrics include traffic volume, healthy endpoint counts, and latency-related metrics. You can set alarms to notify you when latency exceeds thresholds or when endpoints become unhealthy[2][3][6][9].
- Global Accelerator Flow Logs
Flow logs provide detailed records of traffic flowing through the accelerator to endpoints and back to clients. They help troubleshoot reachability and performance issues by showing traffic patterns and potential bottlenecks. Flow logs require setup and use Amazon S3 for storage[2][6][9].
- AWS CloudTrail Logs
CloudTrail records API calls made to Global Accelerator, useful for auditing and troubleshooting configuration changes that might affect latency[2][6][9].
Best Practices for Diagnosing Latency
- Measure latency from the actual client locations to capture real-world performance[1].
- Collect a large number of samples (e.g., at least 1,000 per hour over a day) to capture variability due to traffic peaks and internet congestion[1].
- Ensure endpoints (EC2 instances, ALBs, NLBs, or EIPs) are capable of handling the connection volume to avoid bottlenecks that increase latency[1].
- Use synthetic tests with different tools to validate latency and throughput metrics[1].
- Correlate CloudWatch metrics with flow logs to identify whether latency is due to network issues, endpoint health, or configuration problems[6][9].
By monitoring these metrics and leveraging AWS monitoring tools, you can effectively diagnose and troubleshoot latency issues in AWS Global Accelerator.
Citations:
[1] https://aws.amazon.com/blogs/networking-and-content-delivery/measuring-aws-global-accelerator-performance-and-analyzing-results/
[2] https://docs.aws.amazon.com/global-accelerator/latest/dg/monitoring-global-accelerator.html
[3] https://docs.aws.amazon.com/global-accelerator/latest/dg/cloudwatch-monitoring.html
[4] https://docs.aws.amazon.com/global-accelerator/latest/dg/introduction-speed-comparison-tool.html
[5] https://cloudonaut.io/review-aws-global-accelerator-latency-multi-region-disaster-recovery/
[6] https://docs.aws.amazon.com/global-accelerator/latest/dg/logging-and-monitoring.html
[7] https://digitalcloud.training/aws-global-accelerator/
[8] https://cloud.google.com/spanner/docs/latency-metrics
[9] https://docs.aws.amazon.com/en_us/global-accelerator/latest/dg/logging-and-monitoring.html