Almost every vendor pushing software-defined networking (SDN) of any flavor sells the technology on its efficiency -- networking that is programmable, flexible and can simply do more with less hardware and labor.
The problem is that the efficiency everyone is looking for is operational efficiency -- which at its core means a more favorable ratio of business input versus business output as measured against some benchmark like operating profit. Yet the efficiency that SDN truly delivers is more aptly described in terms of long-term improvement, largely because the time to recoup operational efficiencies is longer than most people realize.
SDN can best be described as a group, or ecosystem, of technologies. The two central elements of SDN strategy are the abstraction of multiple aspects of traditional networking and the opening of APIs for network programmability.
As a group, SDN technologies have some intriguing use cases that will likely lead to a decrease in at least one input to the operational efficiency equation: operational expense (Opex). This will occur initially in the area of change management.
In traditional networks, changes take the form of the manual and the mundane. We move devices from one VLAN or VRF to another, change Quality of Service policies in portions of the network, change circuit speeds and termination points or add and remove servers and services. In most cases, the changes are routine, but you have to touch a lot of devices, so human error tends to creep into the mix. SDN removes much of this risk. SDN controllers, for example, provide deeper visibility into every connection on the network, as well as a central point from which to implement change and record it. This benefit alone advances network change management in light years.
But just to get to that point requires an increase in so many other inputs, from capital expenditure (Capex) on new equipment to other components of Opex, including software design, programming expertise, training and network restructuring. All of this together actually lowers operational efficiency, at least in the near-term.
The question then becomes, if you make these investments, how quickly can you expect to break even? Or, how quickly can you expect to not only recoup the investment money you've put into adopting SDN, but start realizing efficiencies? The honest answer to these questions is, "it depends."
Budget cycles are evaluated annually, but their change is realized over a cycle in several years. The money invested in an upgrade today should be paid back within the first half of that cycle and start gaining efficiencies in terms of a more favorable input/output ratio in the second half.
But that equation is assuming that you are at a place where you are ready to rip and replace most of what you own at the infrastructure level in order to implement SDN. Let's not forget that with SDN, you're generally engaging in a full redesign of your network -- not replacing the core, top-of-rack, and end-of-row switching one for one. Put another way, everything in your network that isn't a server or end-user device is likely to change in an SDN implementation. That's not a small investment, and it significantly changes the balance of the efficiency equation.
I think most people see the possibilities of SDN technology, and they rightly get excited about its potential. But beyond the initial challenge of waiting for the technology to mature (which take up a whole other article), there's also the problem of considering how much you are willing to invest in order to realize the promise of SDN. For many IT pros, SDN is still a solution looking for a problem and won't produce the operational efficiencies touted by vendors any time soon.
About the author:
Teren Bryson is a lifelong professional network engineer, VMware programmer and Unix geek. He is also whiskey taster; long-time practitioner of the art of beating computer and telecommunications systems into submission; brain hacker; student of everything; cancer survivor; lover of stuff that does stuff and a freelance writer. Read his blog at blog.packetqueue.net and follow him on Twitter @SomeClown.
CIOs are taking their time laying SDN plans
Let's get real about SDN!
When building an SDN strategy, think hybrid
- CW500: A roadmap to software-defined everything – Paddy Power Betfair –ComputerWeekly.com
- Software-Defined Networking Could Make Network-as-a-Service a Reality –SearchSecurity.com
- Network Evolution: Software-defined networking in action –SearchNetworking.com
- Computer Weekly – 19 November 2013: Software-defined networking explodes –SearchSecurity.com