Question

A company runs a critical application in a VPC that communicates with an on-premises database over an AWS Direct Connect connection. Users intermittently report slow response times, but the Direct Connect virtual interface metrics and BGP session appear healthy. The network team needs a managed AWS solution that continuously measures round-trip latency and packet loss between the VPC subnets and the on-premises database endpoint, with minimal agent installation and near-real-time CloudWatch metrics for alerting. Which approach best meets these requirements?

Accepted Answer

Configure Amazon CloudWatch Network Monitor with a monitor and probes that specify the VPC source subnets and the on-premises destination IP, publishing latency and packet-loss metrics to CloudWatch.

Answer

Enable VPC Flow Logs with a custom format capturing TCP flags and analyze the logs with Athena to calculate latency and retransmission rates.

Answer

Use Reachability Analyzer between the VPC ENI and the Direct Connect gateway attachment on a scheduled basis to detect the latency source.

Answer

Deploy Amazon CloudWatch Internet Monitor and add the on-premises database prefix as a monitored city-network to track packet loss.