<img alt="" src="https://secure.cope0hear.com/215242.png" style="display:none;">
Onecom blog post

What is the difference between SD-WAN and SDN?


I was presented a few weeks ago with a customer wanting to know all about an SDWAN project. I listened intently whilst the benefits were declared with gusto by the enthusiastic salesman and then gave only the slightest cough before whispering “You are describing SDN not SD-WAN. They are different things..”

I can’t blame anybody for this situation arising except mainly the general body of engineering that decides to spin-up new names and concepts almost with a predilection to confuse and confound the less-technical people they share the Earth with. It has been this way since monks in Medieval monasteries wrote anything interesting in Latin, putting it out of the reach of those who didn’t know their “amo, amas, amat”.

For your sake, dear reader, let me explain what this SD-WAN thing is all about.

Long before anybody had thought about building private networks using technology like MPLS, you could buy a fixed network using physical circuits or something like Frame Relay. Alternatively, you could buy lots of cheap Internet circuits and use a “Virtual Private Network” technology to create encrypted “tunnels” between your offices. The first approach had higher running costs but was reliable, the latter was more expensive to set up, cheaper to run and a bit of a gamble.

Internet bandwidth has got better over the years, the tunnelling technology more efficient and a bundle of other techniques make the whole thing more predictable. As the tunnels are basically just bits of software, you can buy the big boxes at each office and control the whole thing from a central control adjusting how it behaves right down to prioritising a particular application. Thus “SD-WAN” was born. Now I know SD-WAN aficionados are going to say “hang on it does a lot more than that” and they would be right, but for now let’s just be kind to those who have better things to do than read both sides of a datasheet.

Any provider of the underlying connectivity service should be using a different technology set called “Software Defined Networking” or “SDN” which manages how a wide area network is built and configured. SDN has a particular leaning towards MPLS or VPLS as it simplifies the core configuration into a set of services that other layers can manipulate as needed. SDN makes building and changing a customer’s network a lot easier than typing lots of config by hand.

SDN can be used to build the underlying transport system for an SD-WAN but as it can’t control a third-party Internet connection, you lose the benefits of the technology when you try to mix-and-match between providers. SD-WAN can sit quite happily on an SDN-controlled network or be run over the wild west Internet. Or more commonly a combination of the two. So they can complement each other. Or they can run entirely on their own.

So, next time you are in a meeting with an engineer and you don’t understand if they are talking about SDN or SD-WAN, stop them and ask “tu loquere in anglicus”.

James Hickman – CTO Virtual1, 2017