What happened to Software Defined Networking? A while back I wrote a post where I thought 2012 would be the year for Software Defined Network (SDN), and I am really surprised that this technology has not gained greater ground. Now that we are halfway through 2013, I find myself still waiting for adoption of this technology to really take off. With investments from companies like Cisco, IBM, Alcatel, Juniper Networks, Broadcom, Citrix, Dell, Google, HP, Intel, NEC, and Verizon, all of which have current SDN initiatives, SDN will assume a role in IT infrastructure at some point. It just seems like it is going to take a little while longer to catch on.
In case any of you are asking just what SDN is, SDN is a way to expand virtualization into the network stack to create virtualized network for ease of configuration and use in pretty much the same way server and storage are being virtualized. The way SDN works is by introducing a software layer between the bare bones networking components and the admins that configure them. This software layer is basically the control plane or controller which makes the decisions about where traffic will be sent. The data plane which is responsible for moving data forward remains on the physical network hardware. (It is this separation that gives us the network virtualization because you are no longer executing the commands on the hardware itself. The only real difference that I see so far is that SDN is not nearly as far along as its virtual siblings.
The true benefits of SDN are that it is open source technology, and thus is open and vendor neutral, adhering to open standards to be able to operate and control any vendor’s network equipment. In addition, this technology will give administrators greater ability and agility to connect different clouds, application, and/or network devices by taking advantage of the centralized control software for many of the daily tasks they have to perform manually and individually.
So why is it taking too long for SDN to get to the point of mainstream adoption? I really think it is following the adoption path of server virtualization in that companies and enterprises are having a little difficulty wrapping their heads around SDN and how adoption of this technology will save on the overall networking costs and at the same time save them time and money with the overall network operations. It took a while for server virtualization to really take hold as companies took their time to test this virtualization on some low-hanging fruit before moving on. The big difference that I can see between the adoption of server virtualization and the adoption of network virtualization is that most hypervisor vendors hit the market with all sorts of use cases showing how server virtualization will save organizations time and money in their data centers with smaller footprints; there just does not seem to be the same effort by the network virtualization vendors to present some really compelling use cases.
Networking in general is the biggest portion of the datacenter and/or infrastructure that still remains, for the most part, unvirtualized. As the cloud space adoption continues to move at lightning speed, companies and their enterprises will want to federate the networks to be able to move in and out of the cloud. This is where SDN will really shine, by helping to lower project timeframes, which will drive IT networking teams to find more efficient ways to do things than by hand.
SDN is just not a mature technology yet, but most of the major venders have acknowledged that SDN is the future. They have not been able to agree on a common set of interoperability standards despite the fact that SDN is developed based on the open-source heritage. Even if SDN is not fully accepted and developed by all, network virtualization is coming to an enterprise near you in one form or another.