EMAIL SUPPORT

dclessons@dclessons.com

LOCATION

US

VPC Traffic Flows

VPC Traffic Flows

In order to understand the VPC traffic flow, for data to flow, following steps are Nexus follows:

  • VPC traffic flow via ARP Request
  • VPC traffic flow via ARP Reply
  • Data Traffic between Source & Destination

ARP Request:

We will use the following topology to send the traffic from Host A to Host B

  • Let’s suppose Host A With MAC A ( 10.10.10.2 ) want to talk to Host B with MAC B ( 10.10.10.3) and Host A will search for Host B MAC address ins it ARP table and will be a MISS.
  • Now Host A will Send ARP Request with following header details:
    1. DMAC : FFFFF
    2. SMAC : MAC A
    3. SIP: 10.10.10.2
    4. DIP: 10.10.10.3
  • Now when this ARP request will hit to S3 on Eth1/5, Source MAC A will be learned on this interface on its MAC table.
  • Now as this is ARP, HASH algo will be used between DMAC FFFF and SMAC MAC A, and will find one link from Port-channel vPC 10 and let suppose it will select the port E1/1 and ARP will traverse on it.
  • ARP request will hit to Eth1/1 of S1 and puts its entry on MAC table and because of CFS running the Same SMAC MAC A will also be learned on Eth1/2 or port-channel interface on S2.
  • The ARP request will be forwarded to Eth1/2 and Po100 peer-link on S1.
  • Switch S2 and S4 will receive this broadcast and update its MAC table with its respective receiving port-channel interface also S2 will not forward the ARP request to any of vPC member port due to Loop avoidance. Likewise S4 will also not forward this ARP request in vPC 20 but will forward this packet to Eth1/4 where destination Host B resides.

ARP Reply:

  • Now when ARP request is received by Host B , Host B will reply it With ARP Reply and provide its MAC address for which the ARP request has come with following header details:
    1. SMAC : MAC B
    2. DMAC: MAC A
    3. SIP: 10.10.10.3
    4. DIP: 10.10.10.2

GENERAL FAQ

In a vPC traffic flow, communication begins with ARP if the source host does not know the destination MAC address. The source host sends a broadcast ARP request.

The Nexus switch that receives the frame learns the source MAC address on the incoming interface. The ARP broadcast is forwarded across the vPC topology using hashing logic, and MAC information is synchronized between vPC peers through Cisco Fabric Services (CFS).

This ensures both switches maintain consistent MAC tables for proper forwarding.

When an ARP broadcast arrives on a Nexus switch that is part of a vPC, a hashing algorithm selects one active physical link from the vPC port-channel to forward the frame.

This ensures:

• Efficient load distribution

• Prevention of duplicate frame transmission

• Optimal utilization of available links

Hashing is a core mechanism that enables efficient vPC traffic flow.

vPC implements loop-prevention logic.

If a peer switch receives a broadcast frame over the vPC peer-link, it does not forward that frame back out through its local vPC member ports.

This prevents Layer-2 loops while still ensuring that broadcast traffic reaches the intended destination host.

Once the destination host receives the ARP request, it sends an ARP reply containing its MAC address.

The ARP reply is a unicast frame sent back to the original host. During this process, switches update their MAC address tables accordingly, completing the MAC learning process required for proper vPC traffic forwarding.

MAC address synchronization between vPC peers occurs through Cisco Fabric Services (CFS).

When a MAC address is learned on one vPC peer, the information is synchronized with the peer switch. This ensures both switches maintain consistent forwarding tables, allowing traffic to be forwarded correctly even if it enters through different paths.

(Note: MAC learning happens locally first. CFS ensures synchronization. This clarification was required.)

After ARP resolution is complete and MAC addresses are learned, traffic becomes unicast.

Unicast traffic follows the hashing algorithm of the vPC port-channel to select the outgoing physical link. Both vPC peers can actively forward traffic, providing:

• Redundancy

• Load balancing

• Loop-free Layer-2 forwarding

This active-active forwarding model is the key advantage of optimized vPC traffic flow.

Comment

  • RS

    Super Duper Like


  • AB

    Awesome


LEAVE A COMMENT

Please login here to comment.