The debate naturally turned to the need to manage networks more efficiently and it was during this part of the debate that SDN was mentioned. A quick glance around the room showed that – apart from my CTO, James Hickman, who was nodding vigorously in agreement with the speaker – mine was not the only blank face.
Which is quite surprising, since it would appear that in the course of the next 12 months, SDN is going to become big news.
SDN stands for Software Defined Networking. If, like me, this leaves you none the wiser, then the official explanation from James is that it is an application that removes flow control from your network hardware and places it with a software application called a controller, thus allowing network administrators to shape traffic over the whole network from a single, centralised console.
Like virtualisation for the storage and hosting world, SDN will allow network hardware to be provisioned, managed and maintained remotely via a software interface. Obviously this offers some immediate upfront benefits in terms of the day to day management of a network - no more on-site engineers, no carrier change requests, no lead times: everything will be at the fingertips of the network administrator.
But frankly that is hardly going to set the world alight. What really gets James and the plenitude of other techies out there excited is that SDN also allows the network administrator to shape the traffic. Currently, it is the firmware within the hardware which determines to where a packet is forwarded and what priority it has – in fact this currently has to be set manually. SDN removes that ownership from the firmware and gives it to the network administrator. Now this has two implications: firstly, it means businesses will no longer be reliant on the expensive switches which enable the more advanced traffic-shaping, removing the need for such investment. But more importantly, it means a company can be much more agile with the way it moves data around its network because traffic prioritisation can be updated in realtime. It allows much more granular network control with the ability to apply policies at the session, user, device, and application levels – and with network monitoring tools usually incorporated, it means businesses can completely change the structure of their network, without having to touch the hardware.
But as with all new technologies, there are always barriers to adoption, and SDN is no different. The main downside is that SDN is currently nowhere near as advanced as virtualisation for the hosting and storage worlds. This is not a mature product and as such needs to go through the usual birthing pains such as standards ratifications; at the moment, the most advanced products are open source and as such at this time most SDN products are vendor neutral. However Cisco has already brought out its own SDN solution, VMWare acquired SDN developer Nicira over the summer in a multibillion dollar deal and the likes of IBM, Alcatel, Juniper Networks, Citrix, Google, HP, and Intel are all working on their own initiatives, so it’ll be interesting to see how well vendor neutrality fares as the product is commercialised.
The other key barrier of course is in understanding exactly how this is going to make a difference to people’s businesses. How will it deliver its benefits? It currently lacks a champion in the marketplace, and until there is a strong vendor proposition, SDN will continue to be sidelined.
But I don’t believe this will be for long. Hosting virtualisation has paved the way for this product and the efficiencies if nothing else will drive this to adoption.
In short, if SDN isn’t on your roadmap – and your customers’ roadmaps now, it should be. If nothing else, now is a right time to begin to talk to customers about how they might be able to take advantage of the benefits that SDN offers.