Question

A network operations team runs a Transit Gateway connecting 12 VPCs across a region. Users in one VPC report intermittent slow transfers to a shared services VPC. The team suspects a specific TGW attachment is hitting a throughput ceiling but needs data to confirm before escalating. Without deploying any agents, which approach gives the team the fastest way to identify whether that attachment is saturating its bandwidth limit?

Accepted Answer

View the per-attachment BytesIn/BytesOut CloudWatch metrics that Transit Gateway publishes in the AWS/TransitGateway namespace and compare them against the 100 Gbps per-attachment limit

Answer

Enable VPC Flow Logs on every ENI in both VPCs and manually sum byte counts in Athena to estimate throughput per attachment

Answer

Install the CloudWatch agent on EC2 instances in both VPCs to collect NetworkPacketsIn/Out and correlate spikes with the slowdowns

Answer

Run Reachability Analyzer between instances in the two VPCs to confirm the path and measure the achievable bandwidth