What is Shortest Path Bridging and how is it different from traditional computer networks?

Ever since the mankind started to interconnect computers decades ago, IT networks have evolved quite significantly, encompassing everything from small networks with just a couple of switches and a handful of PCs to some extremely complex environments of hundreds of thousands of interconnected computers and devices, spanning over a large number of geographically dispersed locations. And of course, there is the Internet.

This evolution has also brought us a plethora of technologies, protocols and mechanisms, servicing different needs and solving different problems that have surfaced along the way. Some of these protocols and mechanisms are quite complex. But more importantly – they are also interdependent on each other. The end result is that, at one point in time, traditional data networks have become very difficult to build, operate and troubleshoot when something goes wrong. And, true to Murphy’s law, there is always something that can go wrong.

But as it has been since the dawn of time, whenever a new problem arose, a solution wasn’t too far away. In this particular case the solution was called Shortest Path Bridging, or SPB in short.

Shortest Path Bridging, initially specified by the IEEE 802.1aq standard back in 2012 (and later merged into the IEEE 802.1Q standards document), is a computer networking technology that was created with a sole purpose of addressing the challenges of spanning-tree, simplifying the creation and configuration of Ethernet networks, enabling multipath routing of traffic, and importantly: support large-scale virtualization. Personally, I just like to call it Ethernet with a brain. Has a nice ring to it, don’t you think?

What SPB basically did was replace all protocols in the classic network protocol stack with one single protocol: IS-IS, the aforementioned brain (or, more formally, the control plane). IS-IS, a robust, extensible, and well-proven routing protocol, is very much like OSFP (which by the way came later).

Being “the brain”, IS-IS serves two important tasks:

Obviously, Shortest Path Bridging also has a data plane which is taking care of transporting traffic across the network. This is fundamentally Ethernet… however, with some modifications, namely:

One other important property of the SPB fabric network that I want to emphasize is this: traffic will never loop! This allows you to build almost any physical topology, without a single care in the world about loops.

This is due to the property of the control plane: Use of shortest path forwarding. And the data plane: Reverse path forward check. If a switch receives a frame on a port that is on the shortest path back to the source, then everything is OK so it will forward the frame on the port that is on the shortest path to the destination. Otherwise, drop the frame.

Except from this, everything you have ever learned about Ethernet still applies. Nothing new to learn!

Shortest Path Bridging in general, and Fabric Connect specifically – Extreme’s flagship SPB solution for the past 10 years and counting – has a number of sheer technical advantages over other networking solutions, which translate into tangible operational and business benefits. I’ll try to explain them shortly in the paragraphs below.

SPB allows you to build any network topology without having to worry about loops. In fact, the more links and loops you create, the better redundancy and the more capacity!

You can easily scale it (from a few nodes) to support hundreds of network nodes and hundreds of thousands ports. Actually, I know of a certain SPB network that is spanning an entire continent, with four large data centers located in four different cities. SPB networks servicing large hospitals with hundreds of buildings are not uncommon. And then there are SPB networks with just a couple of switches, a few hundred users, and everything in between.

You can easily build secure network zone and it is virtually impossible for a hacker to breach the network itself. This is thanks to the hyper-segmentation and stealth networking capabilities, which you can read about in more detail here and here.

Thousands of virtual networks can be established on top of the physical infrastructure. The standard itself specifies virtual layer2 networks and supports close to 17 million networks which can be stretched to anywhere within the physical infrastructure. Every virtual network is isolated from the other virtual networks – unless you want to allow traffic between some of them. And if you want a really closed network zone, for example for some mission-critical services and devices, a virtual layer2 network can be built without a single IP address. Of course, it can also transport TCP/IP traffic between hosts or any other type of data traffic you may want to or need to support in your network.

This one’s important for everyone who need to support multicast traffic (for video distribution or any other type of multicast traffic). An SPB network means no more need for PIM or any other complicated multicast protocol. All you have to do is ‘enable multicast’. And by the way: it scales beyond any other multicast solution and, in case of failure, it re-converges faster than any other multicast solution. It you’d like to see some real-life examples, check this article about two local government organizations in Europe using Extreme Fabric Connect to support their CCTV systems.

In the next part of the ‘Shortest Path Bridging for Beginners’: