Nexus 3524x, VMware, and our old 2960s stack

[actyler1001]

Hi everyone, first time posting to the forum.  I've managed our VMware environment all built on 1Gb connectivity with our Cisco 2960s switch stack and am very comfortable with the port channel config thus far.  We just signed the PO to bring in a pair of Nexus 3524x and a Pure storage device and I am trying to prep for the deployment.  It seems to me the best way to integrate this in our environment is to create a vPC (which I've not done yet) between the two Nexus switches, then use a vPC port channel to uplink the existing 2960s stack in a redundant fashion..

Am I understanding a vPC port channel correctly in that it will function with another device using standard port channel config..  Both sides don't need to be a vPC in order to function?  just for the initial nexus pair.  I understand the connection between the two nexus switches need to be a pair of 10gb links minimum (in addition to a heart beat link), then member ports can be added as additional port channels to third party or non-vpc aware devices using traditional port-channel technology.

I've also read some concerns around "bridging" in this config...?  We'll be running Per VLAN rapid stp, is there some special config on the Nexus vPC and the 2960s port channel that will prevent blocking?  Any details you can provide or other insight is greatly appreciated!

The Pure storage device will be connected via iSCSI, so I am not planning to use port channeling for it's connection...

[ahmedshoaib · Accepted solution]

Hi;

A virtual PortChannel (vPC) allows links that are physically connected to two different Cisco Nexus Series devices to appear as a single Port-Channel to a third device. 

To configure the vPC first you need to create vPC domain on 2 Nexus switch, requiremnet is you already mentioned in your notes, min 10G link for vPC peer-link & 1G link for keepalive link. you can also used management interface as a keepalive link.

For detail you found the detail information in below links:

As far as a vPC configuration is same as you are doing on 2960 switches the only difference is in Nexus port-channel 1 interface is also part of port-channel with vPC # which will be common on both Nexus:

For e.g Your 2960 is connected to both Nexus on port Ethernet1/1 and you will use vPC 11 & port-channel 11.

Cisco 2960: have the same configuration of port-channel (LACP)

Nexus-1:

feature vpc (which you already configure during vpc domain configuation)

interface port-channel 11
switchport
switchport mode trunk
vpc 11
no shut

interface E1/1
description *** connected to 2960 ***
channel-group 11 force mode active
no shut

Nexus-2:

feature vpc (which you already configure during vpc domain configuation)

interface port-channel 11
switchport
switchport mode trunk
vpc 11
no shut

interface E1/1
description *** connected to 2960 ***
channel-group 11 force mode active
no shut

Note: vPC number and port-channel we are using same, so we know this vpc is belong to which port-channel.

Rapid spanning-tree configuration is also same.

https://supportforums.cisco.com/document/125846/vpc-best-practices-nexus-7000-and-5000

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

Thanks & Best regards;

[actyler1001]

Thanks.  This made a lot more sense after I actually had my hands on the new switches.  We've completed the deployment and all works as you lay out.  Only thing I don't understand is the spanning-tree cost that was assigned to 2Gbps uplink to the existing Cisco 2960.  It gave it a cost of 200 instead of 1000 like you would expect using the long method.  Strange.