Software Defined Networking is the latest buzz hitting the world of networking.
Is SDN little more than hype or is there substance to what's happening?
My personal opinion is SDN has a future and it is delivering real change. Why? For several reasons
1/ The company behind the concept is Google. They couldnt buy OpenFlow technology. They were disappointed with the current generation of equipment so decided to invent SDN and build their own equipment. They are using it today.
2/ The concepts behind OpenFlow are not new. In fact the telecom world embraced the concept of separating control from data more than 30 years ago. Without that concept things like mobile phones would not be possible.
At the moment it is not clear to me what services this concept of separating the data and control planes will enable but it intuitively feels like a more elegant way to build IP networks. I've been running projects to build data centres and networks since around 1998 and my observation is that IP networks have ended up complex and unwieldy. Layers upon layers of boxes have been introduced to overcome shortcomings and deficiencies in protocols or connection models. Roughly every 3 years I used to insist that the designers simplify the network to make it supportable. Although I'm sure OpenFlow will introduce different layers of complexity and support problems in the control plane, it does at least appear to be creating simplicity and elegance in the data plane!
This blog will be a mix of commercial and technical posts. I'm expecting it to be articles sharing my thoughts about OpenFlow as my understanding and insights develop and details of my technical experiments.
I've been involved with OpenFlow (which is one of the most popular SDN technologies) for about 18 months now.
The first experiment I will be sharing is my attempts to build my own OpenFlow switch. It's based on a Raspberry Pi and you too will be able to build your own 4 port OpenFlow switch for about £50 so you can experiment and learn.
No comments:
Post a Comment