Cisco Switch Virtualization

Lessons Discussion
21

Yes, I mean Control Plane.

1 Like
22

Hello, everyone.

I have some questions.

  1. What if a switch in a stack fails?
    Drawing1
    Drawing1948×458 60 KB

If DSW2 here is turned off in the stack, the two links it has connected to ASW1 and ASW2 will go down, right? So will it look like this?

Drawing2
Drawing2415×458 30.2 KB

Although it’s like 1 virtual switch, whether the port is up or down depends on the status of the physical switch, right?

  1. Why don’t any loops occur when stacking switches?
    Drawing3
    Drawing3415×447 35.2 KB

What happens inside the two switches internally? If a broadcast is received, will DSW1 not forward it to DSW2? Or vice-versa.

  1. Are MEC (Multi-Chassis ECs) the result of us using technologies like vPC (on Nexus switches), VSS, Stackwise, etc?

  2. Apart from the Data Plane, the Control/Management planes are centralized, right? If you have 4 switches running OSPF, there will be 1 centralized OSPF process that will be used to build the RIB. The stack won’t send 4 hellos per second for each switch. If you access a switch via SSH and make any new configurations, it will be synchronized with the other switches in the stack.

Thank you.
David

23

Hello David

I recreated the images as you had them, just to make sure we’re ok with copyright rules.

Now, the first thing that I would like to mention is that there seems to be a slight confusion with terminology. The diagrams are labeled “switch stack” but the diagram used is typically used for things like VSS or vPC. In any case, since we’re using the term “stack” I assume it is StackWise that is being used as opposed to other switch virtualization techniques, so I’ll answer your questions in that light.

Yes that is correct. That’s why there are multiple physical links to each ASW. That way you will have active links to the remaining device for continued network operation.

Again, if we’re talking about stackwise, the links you see between the switches don’t act as normal Ethernet connections do. If a broadcast is received on one of the ports of DSW1 for example, it will be forwarded out of all of the ports on DSW1 and DSW2 that belong to the same VLAN. DSW2 won’t forward it back because DSW1 and DSW2 are operating internally like a single switch.

Yes, multi-chassis EtherChannel (MEC), more generally referred to as multi-chassis LAG, allows you to connect downstream devices, such as access switches to two separate upstream devices using a single EtherChannel link that can span multiple devices. This is only achievable if you use vPC, VSS, or StackWise across those multiple switches, making them function as a single logical switch. This design provides resilience and redundancy, making sure to duplicate both the physical links as well as the physical switches to which they are connected.

Yes that’s right. If you have 4 switches stacked together using StackWise, and they are running OSPF, they will run it as if they were one single switch. They will indeed send only 1 hello per interface on a particular network segment, not four, and when you access the stack using SSH, you are indeed configuring the stack as a whole.

I hope this has been helpful!

Laz

24

Hello Laz.

Great, thanks for confirming!

About the “switch stack”. That isn’t something only relevant to Stackwise, is it? Don’t all the other technologies form a stack?

Thank you.
David

25

Hello David

In the context of Cisco, a “stack” refers only to the entity that is created using the Stackwise technology. Other forms of switch virtualization like “domain” or “system” depending on the technology being used.

For example, vPC, the term vPC domain is used to describe the logical grouping of the two participating switches.
In VSS, the two physical chassis are referred to as a virtual switch or VSS system.

I hope this has been helpful!

Laz