Cisco Nexus VPC Peer link lenght

[SebastienWINIARZ63525]

Hello Team,

I have a question about Cisco Nexus VPC Peer link lenght:

Is there a limitation or best practice for the lenght of vpc peer link between 2 Nexus ?

I'd like to create a Nexus Cluster splitted between 2 sites and make vpc peer link through dark fiber, so does someone know any limitation about this design ?

Thanks

Regards.

Sebastien

[Flavio Miranda]

@SebastienWINIARZ63525 

 VPC is logical confg. What  limits the distance is physical. 

https://www.cisco.com/c/dam/en/us/td/docs/switches/datacenter/sw/design/vpc_design/vpc_best_practices_design_guide.pdf

[David Ruess]

Hello,

I would imagine the limit is the cable type you use as fiber length is usually determined by type such as Single-Mode or Multi-mode

-David

[M02@rt37]

Hello @SebastienWINIARZ63525 

When designing a Cisco Nexus vPC peer link over dark fiber between two sites, there are important considerations and best practices to ensure optimal performance and stability. Cisco does not impose a strict maximum distance for vPC peer links, but the design must adhere to latency and performance requirements.

Specifically, the latency between vPC peers should ideally be under 10ms to maintain synchronization and reliable operation. Excessive latency can lead to issues with peer keepalive communication and control-plane updates, potentially causing instability or split-brain scenarios. The vPC peer link itself should be provisioned with sufficient bandwidth, generally at least 10Gbps, to handle traffic replication, unicast, multicast, and inter-switch control communication. It is also critical to ensure the quality of the dark fiber, with minimal packet loss or errors. Transceivers and optical modules compatible with the distance and quality of the fiber must be used, and additional technologies like DWDM can be deployed if the fiber is shared with other services. For reliability, the vPC peer keepalive link must be routed independently of the dark fiber and have minimal latency to detect peer switch status during peer link failures...

https://www.cisco.com/c/dam/en/us/td/docs/switches/datacenter/sw/design/vpc_design/vpc_best_practices_design_guide.pdf

[MHM Cisco World]

Peer link need to be L2 and there is no limitation of length.

Keepalive must be L3  some engineer use WAN for keepalive.

MHM 

[Pavel Tarakanov]

The limitation is in physical possibility to bring up the port (length of cable/fiber and possibility of SFP to deliver enough light power).

On the other hand, I would ask why you need to split VPC cluster between two sites? VPC needed for multi chassis port-channel — to connect endpoints to two different VPC switches. You plan, beside peer-link, to span links from different endpoints between sites?

[SebastienWINIARZ63525]

Hello all, thanks all for answer :).

This design is quite special, for a special client.

We have to build a Etherchannel between one primary site and two DC, with dark fiber, so that's why i asked this question.

All is right with your answers , Thanks !

Sebastien

[vishalbhandari]

Yes, there are limitations and best practices for the length of a vPC peer link between two Nexus switches.

1. Latency: The maximum round-trip latency for vPC peer links should be under 10ms. Higher latency can cause vPC instability and delays in synchronization between the peers.

2. Dark Fiber: Using dark fiber is acceptable, but ensure proper signal integrity, latency, and redundancy.

3. Design Considerations:

   • vPC peer switches should ideally be within the same data center or very close geographically.
   • For site-to-site designs, consider other options like VXLAN or OTV for interconnecting the sites instead of stretching vPC.

A vPC peer link over long distances is not recommended due to potential latency and split-brain risks.