SDN: a vendor’s dream, simplicity’s nightmare

Introduction

You may have heard some rumblings over the last couple of years regarding Software Defined Networking, known better to save you time, as SDN. I’ve listened to countless podcasts, read dozens of blog posts and scoured the Internet trying to make sense of it in that time period and the conclusion I’ve come to is this:

Currently, there isn’t enough joint momentum and focus in the industry around SDN to make this something I need to care about. (12/09/13)

Note that I’ve dated my quote and with a very good reason i.e. I am hopeful that my opinion will change and sooner rather than later.

I liken the recent explosion around SDN to a volcanic eruption. It started off with some gentle rumblings beneath our feet. There were then a series of exciting tremors. Right now, there is a lot of hot air, dangerous fallout and toxic gases, some spectacular fireworks with several ‘oooo’ and ‘ahhh’ moments but it will be some time before things cool down and I’ll want to commit to walking out amongst it.  At that time, I’ll almost certainly blog again about how SDN helps me and my organisation.

In the beginning

When I first started learning about SDN, it was primarily about separating the control plane from the data plane. Centralised, policy based networking. One place to tell your nice GUI what you wanted and a controller that pushed that desire out to your estate. No more logging on to 100 devices and configuring each on a one by one basis. End to end control.

It sounded sweet, made perfect sense to me and was a vast improvement over having a centralised management system that still had to go out to each device that was autonomous and configure them one by one, albeit automatically. It is better still than manually configuring each device which is where many of us are and have been since time began.

Overlays

From intellectual discussions about overlays versus tunnels to ‘cute’ terms like ‘the only way is the overlay’, there is no denying that overlays are a hot topic at the moment. Speak to somebody who is buzzing about SDN and they’ll tell you that you need to make your physical underlay network solid, so you can overlay all of the unicorn goodness etc.

I’ve sat quietly back thinking all the while, why is this new? Surely you always want your underlying network infrastructure to be stable? It allows you to add the overlay networks on top more reliably and helps troubleshooting to boot. Then I saw a tweet from Ethan Banks referring to exactly what I had been thinking:

It seems we are getting excessively excited, not about brand new ways of doing things, but of different ways of doing things that we’ve done before, for many years. Sure, there is some innovative tech hitting the market place but it’s undermined by a deluge of ‘polished turd’ marketing and a lack of standards as each vendor tries to do best by themselves and not the industry.

OK, VMware’s NSX looks like it is willing to take the bull by the horns. Open Daylight looks promising and addresses some of my concerns around a collaborative effort. I do see huge potential with SDN but much more needs to be done.

Bandwagon

It happened with virtualisation, cloud, automation and orchestration. Terms that got molested by sales and marketing folk across the globe. Perhaps ‘cloud’ is the only one in that list that truly got mistreated to the same degree that is happening to SDN right now. It seems that barely a product release goes by without the mention of SDN. It’s actually insulting to me as an IT professional  that these vendors expect me to lap up their wares because I should be caught up in the hype. Let’s calm it down a bit people and start challenging the claims being made. Forget the shiny shiny and let’s start telling, nay, demanding what we want from the vendors.

What question is SDN trying to answer?

This is the key question for me. It’s why I’m not handing out leaflets in my local city centre with the virtues of SDN printed on it, or going to my CTO demanding budget to implement it. It’s also why, until now, I’ve not blogged about it. From clearer beginnings, I am now unsure what exactly SDN is trying to achieve. The waters have been muddied. Or to return to my initial metaphor, there is a lava misinformation regarding SDN.

Summary

I’ve seen the buzz around SDN grow exponentially over the last couple of years but my need hasn’t grown at the same rate. SDN, or rather the industry as a whole, needs to mature considerably before I’ll take it seriously. It needs to become more relevant. Until that time, I’m going to continue building solid networks that run the services that my customers are asking for.

Till the next time.

Please let me know your thoughts!