In a design around fast convergence in case of a link failure, what is the justification for using a point- to-point OSPF network type on the Ethernet links between leaf-and-spine switches on a data c

Although a link failure tears down neighbor relationships on any OSPF network type, this fact describes a universal behavior and does not provide a specific justification for choosing point-to-point over broadcast.

Type 1 Router LSAs are still generated on point-to-point links; it is Type 2 Network LSAs that are eliminated because there is no DR to originate them.

On OSPF broadcast segments, routers must complete a DR/BDR election and wait through the Wait timer before transitioning to Full adjacency, which adds delay to convergence after a link failure. Configuring the point-to-point network type bypasses the DR/BDR election entirely, enabling routers to form Full adjacencies directly and converge faster in a data center leaf-spine fabric.

While eliminating the DR does reduce Network LSA flooding, memory savings are minimal in a modern data center and are not the primary design justification for using point-to-point in a fast-convergence context.

Non-Stop Forwarding relies on graceful restart mechanisms independent of OSPF network type, and the point-to-point type does not specifically enable or require NSF.