1 Reply Latest reply: May 16, 2019 2:45 PM by Micheline RSS

    Peer KAL via Mgmt0 between 2 VDC's on the same chassis

    Veera

      Hi,

       

      Please help me with a brief explanation:

       

      1.  On Nexus 7K, I am planning to form vPC.  Can I create vPC peer keep-Alive link between two different VDC's (let's say VDC 2 to VDC 4) via Mgmt0 interface on the same chassis?. (Only one supervisor engine). If not, please explain why.


      2.  Let's say I have Two chassis. Can I create an Entire vPC feature between VDC 2 of chassis 1 to VDC 3 of chassis 2 using Mgmt0 interface?

       

      3.  Please explain in what scenarios we should not use Mgmt0 interface to from vPC peer keep-Alive link?

       

       

      Pardon me if my questions are invalid

      Micheline_Murphy

       

       

       

      Thank you. 

        • 1. Re: Peer KAL via Mgmt0 between 2 VDC's on the same chassis
          Micheline

          Hello Veera--Your questions are all good questions, so not to worry

           

          1.  On Nexus 7K, I am planning to form vPC.  Can I create vPC peer keep-Alive link between two different VDC's (let's say VDC 2 to VDC 4) via Mgmt0 interface on the same chassis?. (Only one supervisor engine). If not, please explain why.

           

          You can form the peer-keepalive between the management IP of one switch and another, even if "the switches" are VDCs spawned from the same parent switch.  A VDC is basically a virtual switch with a subset of the ports that belonged to the parent, and many of the same characteristics.  One of the things that a VDC gets is a virtual mgmt0 interface, which can be configured with a management IP in the same way that the parent is configured.  In fact, the engineers who designed vPC contemplated the peer-keepalive using the management IPs because when you configure the peer-keepalive, by default it is placed in the management VRF.

           

          Now in a production setting, having a vPC made of VDCs from the same parent switch might not be the best design, because if the parent switch fails, it will bring down both vPC peer switches.  However, in a lab setting, where failure-resistence isn't as critical, using VDCs stretches your lab's capabilities.

           

           

          2.  Let's say I have Two chassis. Can I create an Entire vPC feature between VDC 2 of chassis 1 to VDC 3 of chassis 2 using Mgmt0 interface?

           

          Yes.  VDCs from different parent switches is not a problem, provided that the parent switches are alike enough that the VDCs will pass Type 1 consistency checks.  For example, you would have to double-check the literature if one of your parent switches had only M-series line cards and the other had only F-series line cards.  The Config Guide for Nexus 7000 series switches has a whole long list of caveats to be aware of when configuring vPC... you can look at them here (See Guidelines and Limitations):  https://www.cisco.com/c/en/us/td/docs/switches/datacenter/sw/nx-os/interfaces/configuration/guide/b-Cisco-Nexus-7000-Ser…

           

          3.  Please explain in what scenarios we should not use Mgmt0 interface to from vPC peer keep-Alive link?

          According to the VPC Best Practices, https://www.cisco.com/c/dam/en/us/td/docs/switches/datacenter/sw/design/vpc_design/vpc_best_practices_design_guide.pdf, (at page 30) the recommended peer-keepalive is a dedicated L3 port-channel between the two peers that is at least 2x1Gb.  This design takes advantage of a port-channel's native resilience.

           

          Do these answers make sense to you?  MM