writemem.co.uk

Go Configure!

Understand and Configure Stackwise Virtual (SWV)

6 min read
stack-cisco-switch

In this article I introduce the CONCEPTS, show the USE-CASES, and have a video tutorial on how to CONFIGURE Stackwise Virtual.

Not to be confused with an early 90’s R&B act; SWV is the latest incarnation of virtual stacking from Cisco, called Stackwise Virtual. That’s Cisco, not late 90’s thong-connoisseur ‘Sisqo’, obviously. 

Sisqo not Cisco
Likes it when the beat go.

Sorry, I couldn’t help myself.

Anyway, have no fear the 20+ year old music references are out of the way, stay ‘right here’ for the good stuff. 

What is Stackwise Virtual?

Stackwise Virtual is a way of combining two Cisco chassis into one logical virtual chassis to give resiliency, consolidated management and a shared control plane, while maintaining the ability to utilize both devices for sending and receiving. This means unlike some FHRP’s – we don’t waste ports and uplink bandwidth.

What connectivity is required between SWV members?

To facilitate communication between the stack members and for data destined out of an interface only available on one chassis of the SWV, we use a Stackwise Virtual Link – or SVL. 

On the Cisco 9404 switches that I’ll use in this video tutorial, we could use up to 8x 10Gbps links or 2x 40Gbps links to form the SVL. I’ll be using 2x 10Gbps.

We also should bring up a DAD. A Dual Active Detection link. Its purpose is to ensure the shared control plane in SWV doesn’t become inadvertently active on both chassis. This can happen when there is a problem with SVL connectivity. The DAD is essentially a heartbeat mechanism so it can be a lower bandwidth link, 1Gbps will be fine.

In both cases (for SVL and DAD) you’ll need to use SFP ports on the supervisor module of your device. You cannot configure Stackwise Virtual on a regular line-card interface, believe me – I tried.

The media type for the physical SVL and DAD links can be either copper or fibre. Although most implementations will use fibre due to the greater distance it can carry a signal of sufficient fidelity.

Why use SVW over physical stacking?

Physical stacking suits applications where relatively cheap and very fast intra-stack connectivity is required – but these cheaper pieces of kit tend to have much lower bandwidth uplinks to other parts of the network. Therefore, for a campus core or distribution layer where we we are throwing data around to different parts of the network like a crazed cashier, SWV would be a better solution.

One other limitation of a physical stack is inflexible inter-connectivity. This often involves proprietary stack cables designed for extremely short distances. There’s usually just enough cable to reach a few rack units.

This is not very practical if an organisation (as many do) has two datacentres at opposite ends of a campus. If one DC explodes in a precision asteroid strike, the other DC can take up the slack…or humanity has been destroyed and no-one will be around to care.

SWV makes use of commodity Ethernet connectivity – fibre, or even copper depending on distance – to create its virtual stack. The chassis at each datacentre location acts as one. If physical chassis A meets its fiery demise, chassis B can carry on happily forwarding traffic almost as though nothing had happened. As long as all your connected devices dual homed, and they aren’t also obliterated in the inferno of course!

I use HSRP and it basically does the same thing, right?

The advantages SWV has over first hop redundancy protocols are several-fold:

  • We don’t need to manage SVIs in two places.
  • There’s no need to manage two separate devices at all, in fact.
  • We can utilize all our uplink bandwidth at the same time, rather than waiting for one HSRP member to fail, or manually setting different priorities per vlan and manhandling spanning tree to get it to play nice.
  • Other networking gear can connect to our SWV using Multichassis Etherchannel (MEC) and utilize the forwarding planes of both members. 

So SWV suits all applications then?

Well… not quite. 

There are use-cases for physical stacking and for first hop redundancy protocols, just as there are use cases for Stackwise Virtual.  The implementation needs to suit the circumstances. SWV is a good candidate for a campus core or aggregation layers; where high availability and uplink bandwidth are key considerations.

There are drawbacks; you can’t use SWV on just any old equipment you have kicking around the place. It is a feature of the Catalyst 9k series multilayer switches, as well as 3850’s hardware. Also, a specific “Network Advantage” license is required to utilize the functionality.

Unlike physical stacks and FHRPs you can only have two devices in an SWV. However, there are some big chassis with lots of port density available, so this shouldn’t be problematic in most circumstances. 

You’ll also need to buy two of the same model to bring up an SWV. That’s to say you cannot connect a 9404 and 9407 together and expect them to form a virtual stack. Each device also is required to run the same IOS version – a minimum of IOS XE 16.9.1.

Anything more I need to know before I jump in?

The shared control plane in SWV works by having only one of the chassis – called the ‘active’ node – responsible for the control plane at any one time. The standby node receives L2 & L3 synchronization from the active node so if a failure of the active occurs, the standby can take over.

A heartbeat is sent over the DAD, so even if the SVL fails, the SWV chassis can still talk. This ensures both chassis don’t enter a “dual-active”, sometimes called a “split-brain” scenario. This is when both devices believe themselves to be active for the control plane, causing route flapping for connected devices.

Stackwise Virtual – The “Gotchas”.

The first gotcha is that SWV uses VLAN 4094 for its intra-stack communication. So if you were planning to use 4094 in the same Layer 2 domain, well, you can’t. It’s time to migrate to another VLAN unfortunately.

Secondly, if you have purchased devices with dual supervisor cards…perhaps you’ve inherited the kit…  then you’re unable to use both at the same time. One of the SUPs on each chassis will be a cold standby.

Configure Stackwise Virtual

For those that have been around enterprise class Cisco gear for a while, you’ve probably heard of SWV’s predecessor VSS – and if you’ve configured VSS then great news! There’s not too much here that you haven’t seen before. 

Take a look at my video below demonstrating how to configure Stackwise Virtual on Cisco 9400 Series chassis – in 7 easy steps:

Configure SWV in 7 (Easy) Steps!

FIN.

I hope that this article has proved valuable to you in understanding and configuring Stackwise Virtual. Check back soon for more news, reviews, and tutorials!