passdrill
AWS SAA-C03 · VPC & Networking · Card 029/038 medium

A private subnet's outbound traffic through a single NAT gateway is failing intermittently for connections to one specific busy third-party API endpoint, while traffic to other destinations is unaffected. Monitoring shows the NAT gateway's IP address is hitting its concurrent-connection ceiling for that one destination. What AWS-recommended fix directly raises this specific ceiling for the existing NAT gateway, without changing which subnet or route table the traffic uses?

  1. Associate additional secondary Elastic IP addresses with the same NAT gateway; each IP address on a NAT gateway supports its own separate pool of concurrent connections to a given unique destination (a specific destination IP, port, and protocol combination), so more addresses raise the effective ceiling
  2. Increase the NAT gateway's instance type, since NAT gateways run on a selectable EC2 instance type and larger instance types support proportionally more concurrent connections per destination
  3. Enable VPC Flow Logs on the NAT gateway's network interface, which automatically raises its per-destination connection ceiling once AWS detects sustained legitimate traffic
  4. Change the route table's default route from the NAT gateway to an internet gateway, since internet gateways have no per-destination connection ceiling
Next card → Shuffle