A second AWS Direct Connect connection is added to an existing setup. A resiliency test shows a failover event causes a 90-second traffic interruption before the backup connection carries traffic. What change will shorten failover time?

Configuring Bidirectional Forwarding Detection (BFD) on the on-premises router will significantly shorten failover time. BFD is a lightweight protocol that quickly detects failures in the forwarding path between two routers. When integrated with BGP, BFD can detect a link failure in milliseconds, allowing BGP to converge much faster than relying solely on BGP's default hello and hold timers. This rapid detection enables the backup Direct Connect connection to take over traffic almost immediately, reducing the 90-second interruption. Decreasing the BGP hello timer to 5 seconds and decreasing the BGP hold-down timer to 5 seconds would improve failover but not as effectively as BFD. BGP timers are designed for routing protocol convergence, not rapid link failure detection. While reducing them helps, they still operate on a slower timescale than BFD. Adding a VPN connection provides another layer of redundancy but doesn't inherently shorten the failover time for the existing Direct Connect setup; it's a separate connectivity solution.