Cisco Switch Virtualization

In a previous lesson I explained about campus network design and how we use different layers and “switch blocks” to create a hierarchical design that has redundant links.

You also learned in the spanning-tree lesson how spanning-tree creates a loop-free topology by blocking some of the redundant links. The “thing” with spanning-tree is that we have a loop-free topology, we have redundancy but we can’t use all the redundant links for forwarding. Here’s an illustration to visualize this:

Spanning-tree access-layer blocked ports

The dashed lines are layer 2 links. Spanning-tree will block two of these links to create a loop-free topology. Another issue with this topology is that we do have redundancy in the distribution (and core) layer but we don’t have redundancy in the access layer.

When one of the distribution layer switches fails, the other one can take over. We don’t have this luxury in the access layer…when either of the switches fails then the other one can’t take over.

One way of solving this problem is to create a logical switch. Cisco switches offer some technologies to convert two or more physical switches into a single logical switch, it will look like this:

access layer logical switch

A1 and A2 are two physical switches but they are combined into a single logical switch. The distribution layer switches think that they are connected to one access layer switch. The uplink pairs to each distribution layer switch can be combined into an Etherchannel.

When the link between D1 and D2 is a layer 2 link, spanning-tree will still have to block one of the etherchannels.

We can improve this topology by doing the same thing in the distribution layer, combining the two physical distribution layer switches into a single logical switch:

access distribution layer logical switch

The two distribution layer switches are now combined into a single logical switch. The four links between the switch pairs can be combined into a single etherchannel. Since we now have a single link between the two logical switches, spanning-tree doesn’t have to block anything. Normally we can’t create an etherchannel that spans multiple physical switches. By creating logical switches, this is no problem. An etherchannel like this between multiple physical switches is also called Multi-Chassis Etherchannel.

Combining multiple physical switches into logical switches makes our network topology a LOT simpler, here’s a “before and after” example:

We're Sorry, Full Content Access is for Members Only...

If you like to keep on reading, Become a Member Now! Here is why:

  • Learn any CCNA, CCNP and CCIE R&S Topic. Explained As Simple As Possible.
  • Try for Just $1. The Best Dollar You've Ever Spent on Your Cisco Career!
  • Full Access to our 654 Lessons. More Lessons Added Every Week!
  • Content created by Rene Molenaar (CCIE #41726)

540 Sign Ups in the last 30 days

satisfaction-guaranteed
100% Satisfaction Guaranteed!
You may cancel your monthly membership at any time.
No Questions Asked!

Tags: , , ,


Forum Replies

  1. Hi Zaman,

    Switch virtualization is great for redundancy but it also simplifies your switching a bit. You’ll have less spanning-tree to deal with since all switches in the stack are one “logical” switch.

    What you use on the access layer really depends on your design and your budget. Normally we do have redundant connections from the access layer to the distribution layer.

    Rene

  2. Hi @the.prince.of.nyinyi,

    That’s something I will cover once I start with my Nexus material. Here’s quick example though:

    https://www.cisco.com/c/dam/en/us/products/collateral/switches/nexus-7000-series-switches/configuration_guide_c07-543563.doc/_jcr_content/renditions/configuration_guide_c07-543563-2.jpg

    With VPC, you can configure a port channel on those Nexus 2000s with the physical links that connect to the 5000 switches. From the 2000’s perspective, it will appear as if it’s connected to a single switch.

  3. Hi Rene,
    this is very useful article, and it’s very important topic…thanks

    regards,

  4. Rene’,

    In part of your discussion, you note that “When one of the distribution layer switches fails, the other one can take over. We don’t have this luxury in the access layer…when either of the switches fails then the other one can’t take over. One way of solving this problem is to create a logical switch.”

    From what I’ve seen, using stackwise still doesn’t provide redundancy at the access layer. Let’s say that the third switch in the stack dies. All users that are plugged into that third switch are now dead in the water, right?. I don’t see how stackwise is

    ... Continue reading in our forum

7 more replies! Ask a question or join the discussion by visiting our Community Forum