Ronny Lam

about://tech

Don't Let SDN Be a Goal on It's Self

Software Defined Networking is hot, so much can be said. While the definition is still not exactly clear, every self-respecting networking company is jumping on the bandwagon. The question is why? What problem are they trying to solve? What is the market asking for? I think a quote from the Plexxi Whitepaper sums it up very well…

Unfortunately most SDN discussions quickly dissolve into architectural and implementation dissertations about decoupling forwarding and control plane functionality and providing centralized programming interfaces. By going directly to architectural issues, we’ve put the cart before the horse, and lost sight of the higher-level functional goals.

Which is true, amongst the people that build these tools and the networking community. However, (large) enterprises are still not considering SDN. They talk about it, but don’t have a clue about what it means. What is it that these companies want? The network has to support their business goals and is not a goal on its self. Most of these companies still have tons of legacy hardware.

And this is where the trick comes in. OpenFlow, or any other controller-based network, requires you to invest in new hardware. And to overcome that other companies sell overlay networks to run SDN-like applications on an existing network. But who is thinking of the legacy network? Nobody! It’s either buying new hardware or hiding the existing network, which still has to be managed.

So, what is the market asking for? The market for Datacenter networks is clear. But for telecom networks, MSO’s and (large) enterprises that goal still has to be defined. Instead of diving into architectural issues we have to support their business drivers whereby stability of the network is number one and time to market is a close number two.

Let SDN not be a goal on it’s self but let it support what the business really needs.