Stacking Network Switches: Why and Why Not

By Andrea Mauro, Contributor
Share Post

What are Stackable Switches?
In networking, the term “stack” (or stackable) refers to a group of physical switches that have been cabled and grouped in one single logical switch. Over the years, stacking features have evolved from a premium (and costly feature) to a core capability of almost all enterprise-grade switches (and also in several SMB models).

A stack of switches (in this example Aruba 3810 Switch Series)

It’s the opposite approach of a modular switch, where you have a single physical chassis with several slots and modules to grow your switch, used typically, at least in the past, in core switches.

A modular switch (in this example: Aruba 8400 Switch Series)

Both can provide a single management and control plane or at least a single configurable logical switch, with some kind of redundancy if you lose a physical switch or a module.

Having a single logical switch, with better reliability, makes it easy to translate the logical network topology in physical topology.

What are Stacking Technologies?
In stackable switches, the stack is usually built with cables that connect all the switches in a specific topology.

Those cables are connected to specific posts of the switches, depending on the type of stacking:

  • Backplane stacking (BPS), where specific stacking modules (usually on the back of the switch) are used with specific cables (depending on the vendor).
  • Front-plane stacking (FPS)-VSF, where usually are used standard Ethernet ports to build the stack, using standard Ethernet cables.

The stacking topology also define the resiliency of the stacked solution, you can have typically different kind of cabling options (depending on the switch vendor and models):

  • Daisy chain or bus is not usually used because it does not provide resiliency
  • Ring or redundant dual ring provide resiliency, but with more than two switches the packet paths can be not optimal
  • Mesh or full mesh provide higher resiliency and also optimal packet paths

For example, the Aruba 3810 Switch Series uses a dedicated stack module (on the back) and supports all those topologies (the ring topology only with a single ring).

In the ring topology you can have up to 10 stack members:

Aruba 3810 Switch Series stacking: Ring topology

In the mesh topology you can have up to five stack members:

Aruba 3810 Switch Series stacking: Mesh topology

Why Use Stacking?
One of the major benefits of using stacking (depending on the vendor) is the logical switch view with a single management interface, which  makes the management and operational tasks very easy.

It also provides link aggregation capability between ports of different physical switches in the same stack, providing better bandwidth and resiliency for the downstream links, and simplifying network design implementation, where “multiple cables” across switches are just one single logical link (using LAG, LACP, EtherChannel or any link aggregation solutions).

Compared to the modular switch option, stackable switches provide a less expensive option (especially for SMB use cases), but with similar scalability and usually with better flexibility. Resiliency and performance can be different (worse or better) depending on the implementation.

With regards to flexibility, you can usually mix a combination of different port speed and media types, but also mix different models of switches with also different capabilities (for example, some switches with PoE functions).

Talking about performance, stacking switches doesn't not necessary means increase the performance. This depends on the stacking cables' bandwidths and the stacking topology.

Why Shouldn’t You Use Stacking?
The stackable switch market is very mature and relatively stable. However, each vendor adds its unique set of features and functionalities. Different vendors utilize different types of connectors, cables and software for their stackable switches. This causes requirements to use the same product line of switches to take advantage of stacking (not necessarily the same model, because, for example, in Aruba 3810 Switch Series you can mix different models in the same stack).

And there are other potential disadvantages when you use stacked switches:

  • Performance: For SMB use cases, the stack ports and cable speed are enough to provide high bandwidth and low latency. But when speed increases or the stack expands (unless you use a mesh topology) you may increase the latency and decrease the overall performance.
  • Resiliency: Depending on the stacking topology, if you have some faults your overall stack may not be operating correctly anymore. So be sure to choose the best topology and ensure higher resiliency on each stack member. For example, using dual power supplies to ensure hardware redundancy. The single management or control plane may also reduce the overall resiliency, but the problem is similar also on modular switches.
  • Manageability: The single management interface is great, but there are also some drawbacks. First, expanding an existing stack could cause a service disruption for an extended period, such as when all the switches are rebooted to add a stack member or from a power failure. Second, removing a switch from a stack could be tricky or require a complex process. Last but not least, upgrading the firmware on all the stack members, in most cases, requires a complete reboot of all the switches.

Stacking Evolution
To increase the resiliency of stacked switches, there are different solutions based on the concept of a "virtual chassis" with separated management and control planes. Usually, those solutions are implemented on high-end switch models.

Each vendor has its proprietary solution, but for example, on the Aruba 8320 Switch Series (or the Aruba 8325 Switch Series) you can use the Aruba Virtual Switching Extension (VSX).

Related Content
Aruba Switches Series Overview

Configuring an Aruba Switch Stack on Aruba Central

Aruba OS Switch Stacking for Airheads

Blog: Network Security on Modern Switches