Which options do you investigate first when designing fast network convergence?

Routing protocol database size affects SPF computation time, which is a later convergence stage that is irrelevant until a failure has already been detected.

Interface MTU affects packet delivery and adjacency stability but is not a primary determinant of how quickly the control plane identifies and reacts to a topology change.

Link speed influences throughput and physical propagation delay but does not govern how fast the control plane detects a link or node failure and triggers reconvergence.

The Layer 3 failure detection mechanism - such as BFD, routing protocol hello/dead timer tuning, or interface carrier-delay settings - controls how quickly a failure event is signaled to the routing process; until a failure is detected, SPF calculation, route installation, and traffic rerouting cannot begin, making detection latency the single most critical variable to investigate first in any fast-convergence design.