In one of the spine and leaf designs, VMware vSphere is used including VMware NSX. OSPF is used for routing protocol across the spine and leaf topology. What do you need to consider in this design?…

Question

Options

• AAdding an anycast loopback interface on each VSX pair
• BUsing a L2 MTU of 9198 on leaf-to-spine links
• CUsing an IP MTU of 9198 on leaf-to-spine links
• DEnabling OSPF in passive interface mode on the NSX transport VLAN

• A
  28% (11)
• B
  72% (28)

Explanation

In a VMware NSX spine-and-leaf design, NSX uses GENEVE encapsulation for overlay (TEP) traffic, which adds significant packet overhead - requiring jumbo frame support at both the L2 (interface) and L3 (IP) levels. Setting both to 9198 bytes on leaf-to-spine links (B and C) ensures encapsulated packets traverse the fabric without fragmentation, which is an HPE/Aruba-specific dual-layer MTU requirement. Enabling OSPF passive interface on the NSX transport VLAN (D) is critical because that VLAN carries TEP (Tunnel Endpoint) traffic between ESXi hosts - passive mode still advertises the network into OSPF but suppresses Hello packets, preventing OSPF from accidentally attempting adjacencies with ESXi hosts or other non-router endpoints.

Why A is wrong: Anycast loopback interfaces on VSX pairs are relevant in EVPN/BGP-based fabric designs (for distributed gateway scenarios), not in OSPF-only spine-and-leaf designs with NSX. It's a tempting distractor because VSX is common in these topologies, but it addresses a different design pattern.

Memory tip: Think "Big frames, quiet transport" - you need big MTU values (9198) at both L2 and L3 to carry NSX overhead, and you keep OSPF quiet (passive) on the transport VLAN so it doesn't gossip with ESXi hosts that can't speak OSPF.

No community discussion yet for this question.