A service provider wants to maximize forwarding memory, routing memory, and CPU resources on PE devices for enterprise customers that have very few sites (two to three) and a large number of prefixes

Question

Options

• AVPWS
• BVPLS
• CL3VPNs (RFC 2547-based)
• DInterAS L3VPN

• A
  59% (16)
• B
  7% (2)
• C
  22% (6)
• D
  11% (3)

Explanation

The goal is to preserve (maximize available) PE device resources - forwarding memory, routing memory, and CPU - for customers who have few sites but thousands of prefixes. In an L3VPN (RFC 2547), the PE must maintain a VRF routing table containing all customer prefixes. With thousands of prefixes per customer, this heavily consumes PE memory and CPU. VPWS (Virtual Private Wire Service) is a Layer 2 point-to-point pseudowire service. The PE device simply performs label switching between two endpoints - it never learns or stores any customer IP prefixes. Since these customers only have 2–3 sites, VPWS fits perfectly: you need just a small number of pseudowires, and the PE remains completely unaware of the customer's IP topology regardless of how many thousands of prefixes exist inside the customer network. VPLS would similarly offload IP routing but requires the PE to maintain MAC address tables per service instance. L3VPNs and InterAS L3VPNs are eliminated because they require PE devices to hold all customer routes in VRFs, which is exactly what exhausts resources at scale.

No community discussion yet for this question.