The SDN misunderstanding

Software-defined networking is not a networking paradigm shift, and it's not the holy grail of networking.

Read time 4min 30sec

One of the more surprising aspects of 2013 has been the rush by IT industry professionals to join the SDN (software-defined networking) bandwagon.

Investors are buying into start-up companies which have proliferated. They are encouraging early adopters to join the party. All this without a clear understanding of SDN, its true role in the industry, or its future direction.

As a result, some of the largest vendor organisations have been obliged to react to what they perceive to be a demand for a technology that basically allows users to substitute some hardware functions in server switches with software, while emphasising automation.

This has led to a degree of confusion, because as it transpires, each vendor seems to have its own take on SDN. And varying standards brings with it the danger of an SDN adopter being tied to one supplier for the foreseeable future.

For the record

I'm on record as saying that SDN is not going to be everything it's made out to be - certainly not the 'next big thing' in networking. And I've pointed out that SDN, in other guises, has been seen before. Moreover, its goals of networking simplicity and ease of use are by no means new.

So, what is the future for SDN?

Rather than the 'holy grail' of networking - as many proprietary SDN vendor solutions are purported to be - I see SDN simply evolving into another element in the colourful, ever-expanding networking mosaic.

Why? Because it will take far too long for a common denominator to be found linking the various 'flavours' of SDN now being touted by vendors whose vested interests will surely trump any desire for co-operation.

In addition, there are too many industry associations keen to be seen driving the SDN bus - including OpenFlow and the Open Daylight consortium - for a single destination point to be clearly defined.

It could be 10 years before an accepted SDN version is ratified, and in this period, another new technology is sure to make its presence felt in a market keen to embrace new ideas.

Celebrity status

This begs the question: why has SDN garnered so must attention so quickly?

Accepted, there is a demand for a solution that delivers increased flexibility and manageability in the data centre, but this has come from a small segment of the market - operators of exceptionally large networks.

I see SDN simply evolving into another element in the colourful, ever-expanding networking mosaic.

Misguided mainstream demand stems from the 'hype' surrounding cloud computing and virtualisation with which SDN has been closely associated.

This is because SDN decouples the network control system that defines where data traffic is sent - the control plane - from the underlying system that forwards traffic to the selected destination - the data plane. It is thus seen as an interface between cloud computing and the network. Some industry watchers predict SDN 'as a technology' will replace the network infrastructure as it is known today.

While these conjectures are largely irrelevant, SDN is sure to find its niche in a less conspicuous role.

It could become a 'killer app' with relevance in future environments, where more powerful, commoditised network switches will be readily available and more easily programmed by downstream APIs (application programming interfaces).

For example, a network virtualisation/SDN solution may well be developed for the data centre. It will focus on managing the switching of virtual ports, allowing servers to 'speak to one another' across virtualised, rather than physical, paths.

Another SDN application could be for data analysis. Here, different flows of data in the network could be directed by the SDN app, allowing analysts more visibility and control.

These benefits would be most applicable to large Internet 'pipes' such as those found in major corporates or research facilities, where some form of SDN would ideally do duty at the extreme edge of the network.

In all probability, the IT industry could well gain access to a suite of future SDN killer apps, each with a clearly defined focus area of its own.

One of the more important reasons for SDN's apparent popularity among operators of large networks is its promise to cut operating expenses by reducing data centre staff complements. In theory, this is a good idea.

However, the remaining data centre staff would have to be highly skilled (and highly paid). In Africa, there is limited access to this level of expertise.

In summary, expect SDN to play a role in large corporate and public sector networks of the future, but it will be limited to specific parts of the network. SDN will not be the ubiquitous solution many view it to be.

The final word must go to Reuters News Agency writer Sayantani Ghosh who, following a recent survey, said participants reported seeing an actual SDN deployment "as often as they've seen Bigfoot, Elvis or the Loch Ness Monster".

Andy Robb

Technology specialist at Duxbury Networking.

In his role as CTO, Andy Robb is Duxbury Networking's chief technologist and technical advisor, responsible for the company's strategic technical direction. Robb oversees quality of service delivery and product management. He holds a number of industry product-related qualifications as well as continuing with further tertiary qualifications. Prior to becoming CTO, Robb held a variety of positions at Duxbury Networking, including technical manager, product manager and senior systems engineer. He has been with Duxbury Networking since 2000.

See also