In an EVPN-VXLAN data center, what would be a reason to consider using discrete border leafs rather than placing border roles onto the spines?

Question

Options

• Ato provide for differentiated oversubscription ratios for traffic into and out of the data center
• Bto eliminate multicast traffic in and out of the fabric
• Cto allow for provisioningACLs between the data center and the WAN
• Dto provide specific devices designed for handlingtraffic burstiness on the WAN

• A
  73% (19)
• B
  4% (1)
• C
  15% (4)
• D
  8% (2)

Explanation

Using discrete border leafs lets network architects choose hardware with different oversubscription ratios than the spines. Spines are typically sized and oversubscribed for predictable east-west (intra-fabric) traffic, while WAN-facing traffic is north-south with different volume profiles - border leafs can be independently sized with port densities and oversubscription ratios appropriate for that external role, without forcing the spines to compromise their fabric-optimized design.

Why the distractors are wrong:

• B is incorrect because the choice between spine-border vs. dedicated border leaf roles has no bearing on multicast usage in the fabric - BUM traffic handling is a separate design decision.
• C is incorrect because ACLs between the DC and WAN can be provisioned regardless of whether the border role lives on spines or dedicated border leafs.
• D is a tempting near-miss - WAN burstiness is a real concern - but it's a consequence of the underlying reason (different traffic profiles require different hardware sizing), making A the more precise and complete answer.

Memory tip: Think of spines as the "internal highway" and border leafs as the "toll plaza" - you wouldn't redesign your whole highway system to handle toll booth traffic; you add a dedicated structure sized for that specific load.

No community discussion yet for this question.