You are adding EVPN VXLAN to an L3 spine and leaf topology that uses OSPF for the underlay. You plan to use a single BGP AS. Which design for EVPN neighbors is correct?

Question

Options

• ALeaf switches should establish iBGP sessions with other leaf switches using their VXLAN loopback
• BThe border leaf switches should act as route reflectors and specify the leaf switches' VXLAN
• CAt least one spine switch should act as route reflector and specify the leaf switches' unique
• DLeaf switches should establish iBGP sessions with spine switches on their actual P2P link IP

• A
  5% (1)
• B
  14% (3)
• C
  71% (15)
• D
  10% (2)

Explanation

In an iBGP EVPN VXLAN fabric with a single AS, spine switches are the natural Route Reflectors because they already sit at the center of the topology - every leaf has an uplink to every spine, making spines the ideal hub for reflecting EVPN routes between leaves using loopback IPs as BGP peering addresses.

Why the distractors are wrong:

• A - Leaf-to-leaf iBGP full mesh is unscalable: N leaves require N×(N-1)/2 sessions, and EVPN fabrics are designed explicitly to avoid this with RRs.
• B - Border leaves handle external connectivity (DCI, WAN hand-offs), not internal route reflection; placing RR responsibility on the border leaf creates a suboptimal, asymmetric design.
• D - BGP sessions should peer over loopbacks, not P2P link IPs. If a P2P link fails, the BGP session tears down even if an alternate underlay path exists; loopbacks provide session resilience through OSPF's multipath.

Memory tip: Think "Spine = RR" - the spine is already the physical hub of your fabric, so let it be the BGP hub too. Loopbacks for BGP, P2P links for OSPF - keep each protocol in its lane.

No community discussion yet for this question.