Software defined networking: the future of wholesale telecoms?

This month, engineers from some of the world’s biggest carriers and equipment providers will converge on Indiana University’s Bloomington campus for something called a ‘Plug Fest’.

Don’t be put off by the rather incongruous sounding nature of the event. It promises to be the hottest show in town.

The meeting, which is organised by the Open Networking Foundation (ONF), is an ‘invitation-only’ gathering at which 20 or so developers will showcase their blue-prints for a new generation of computer-driven networks that will revolutionise the way carriers operate from the bottom up.

Behind the closed doors of IU’s state-of-the-art InCNTRE laboratory, attendees lucky enough to snag a pass will see how far their peers have got in developing systems and hardware to embrace a burgeoning initiative called software defined networking (SDN).

Expectations are high. The foundation, which is backed by Verizon, France Telecom, Deutsche Telekom, NTT, Microsoft and Google (among others) believes that SDN will do for the telecoms industry what the microprocessor did for computing, heralding a paradigm shift in the way networks are designed and run.

And to judge by the number of carriers and service providers that have already signed up to the foundation – 78 at the last count – a powerful and growing industry consensus is fast mobilising behind the forum.

Hindsight is, of course, a beautiful thing. But it turns out that network architects have been running down a blind alley these past few years. Considerable time and money has gone into developing highly intelligent routers and switches capable of carrying a huge amount of processing power and proprietary software.

This cutting-edge equipment might be able to think very effectively for itself but in the rush to innovate, a simple premise has somehow been forgotten – because these routers and switches are designed to work independently of one another, most of that brainpower ultimately languishes, unused.

SDN changes the game by taking all that autonomous intelligence off the network and handing it over to a single all-seeing, all-powerful computer. In techno-speak, this is known as separating out the data plane, which is responsible for shunting packets of information on their way along the network, from the control plane, which decides when and where that data goes next.

Conceptually, it might sound like a relatively small step. But in practice, handing control of a network over to a programmable computer potentially opens up a once-in-a-generation opportunity for carriers to re-invent their business models.

To some, who face an uncertain outlook plagued by flat-lining revenues and surging costs, SDN might offer their best hope of survival. To others trapped in a commoditised future providing dumb pipes to content providers, the advent of SDN offers a chance to develop high margin services, finely tuned to individual customer needs.

As organisers of the ONF’s Plug Fest will no doubt be keen to show, the technology behind SDN is largely proven. “The genie’s already out of the bottle”, affirms Stuart Elby, vice president of network architecture and chief technologist at Verizon. “We already know that the technology works. The issue for us now is how quickly we can deploy it and what financial benefits it can deliver.”

Indeed, one of the great expectations of SDN is that it will generate huge cost savings by helping carriers optimise traffic flow. It is estimated that some networks currently run at just 40-50% optimisation and that software-driven applications might help improve this to nearer 90%.

As Sanjeev Mervana, senior director of marketing for products and solutions at Cisco explains, carriers are sitting on a vast pool of proprietary information that might help them slash costs and bolster efficiencies. SDN is the key that could help them unlock that data – and more importantly, monetise it: “It may be a long time before carriers chance across a better opportunity”, he says.

This summer, Google thrust the concept of SDN centre stage when the search engine announced that it had moved all its backbone traffic, one of the largest networks in the world, entirely over to a software-driven architecture.

Urs Hölzle, the internet giant’s legendary technology guru, revealed that it took less than two years to make the jump, adding that the company had been forced to design and make its own gear in order to speed up the deployment. That alone was enough to send a tremor of panic through some incumbent equipment vendors and both Cisco and Juniper have been at great pains to reassure Wall Street analysts that they continue to have a business model in a post-SDN world.

Hölzle would not be drawn on details, but he confirmed that Google had already seen significant cost savings after making the switch and that his engineers were now focussing their attention on Google’s customer-facing architecture – the network that delivers applications such as YouTube and Maps to internet users.

Eventually, Hölzle says, the company hopes to be able to use an SDN to establish temporary virtual private networks to allow groups of people to chat via a video link using Google+, eliminating performance issues such as bandwidth limitations. But that’s just one potential application.

The beauty of a software-dependent business model is that, ultimately, you are limited only by the imagination of the people you bring in to write the code. A raft of applications become possible in an SDN environment.

Traffic flow, for instance, becomes controllable at the touch of button. Bandwidth can be turned up or down at the flick of a switch. A number of virtual networks can be run simultaneously on just one physical network of switches and routers. The computer can be pre-programmed to respond to any number of possible disaster scenarios.

In short, the network can be reconfigured an almost limitless number of ways to match specific customer demands in real time. And because software is considerably easier to upgrade than hardware, new applications can be designed, developed and deployed within a matter of weeks rather than years.

As Verizon’s Elby explains, SDN hands the power of innovation back to the carriers: “If we want to launch a new service today, we have to bring on board four or five vendors and specify exactly what special features we are looking to incorporate into the network.”

The resulting hardware can take a long time to build, even longer to test, and longer still to integrate, which is why carriers take years to ensure that there is a solid business case to answer before pressing the button. If all that innovation subsequently shifts to a software plane, the dynamic changes considerably. “It gives us flexibility and speed, and it reduces the cost of launching a new service, which ultimately allows us to take a few more chances,” Elby says.

And that’s just scratching the surface. Eventually, an SDN might go on to spawn a single killer application that could define the future of wholesale telecoms, or it might foster thousands of smaller applications that carriers and service providers never knew they couldn’t do without until they were brought to life.

“As a carrier, it’s very rare that you face a technological disruption that can have such huge ramifications across the whole ecosystem of your business”, says Elby. “SDN might lead to the emergence of new carriers with entirely new business models, or it might sound the death knell for those who find they can’t adapt quickly enough.”

There is but one caveat: you need the controller to be able to talk to all the devices on a network and for that, you must devise a new computer language or protocol. And this is where the ONF is most vocal.

Supported by luminaries such as Hölzle, a Google fellow and one of the first 10 employees to work at the search company, Najam Ahmad, director of technical operations at Facebook and Scott Shenker, who backed Nicira, one of the most successful start-ups in the SDN arena, the foundation is determined to standardise its own OpenFlow protocol to serve as a platform for future research and development.

Dan Pitt, the ebullient head of the ONF, puts it rather more succinctly: “We want to change how virtually every company with a network operates.” Five years from now, he adds, carriers should not be asking themselves how to make SDN work, but how they ever managed to survive without it.

In this respect, this month’s Plug Fest is not just a hot trade show, but an opportunity to prove that emerging equipment providers such as Big Switch and Arista Networks can develop interoperable systems and devices using the ONF’s OpenFlow language.

Some industry insiders say that the ONF’s protocol is not necessarily advancing at the same speed as SDN and there are clearly some gaps in the software specification that worry many carriers.

In particular, the programming around the controller itself needs to be more robust. If you are going to put a computer at the heart of your network, you need to be assured that it will have five-9’s reliability and that the control function can switch seamlessly to a back-up system in the event of a problem.

There are also issues around security and the extent to which the control function might be prone to cyber attack. The latest incarnation of OpenFlow, Version 1.3, will address at least some of these concerns although few equipment vendors have yet to launch products using the latest coding.

In reality, SDN ecosystems should never be reliant on a single protocol: “It certainly has an advantage here and now but SDN won’t live or die on the success of OpenFlow”, says Verizon’s Elby. “So long as the SDN framework is good enough then it should be able to adapt to a better protocol if one comes along. You shouldn’t need to have to change the whole architecture, just the protocol.”

But arguably, the greatest challenge facing carriers is not so much the technicalities of implementing OpenFlow, but the huge mind shift that operators will be required to make to move over to a software-driven network ecosystem.

The whole idea of SDN is to virtualise the network and make it look more like a data centre. That’s a laudable aim – data centres are run very efficiently and carriers will undoubtedly save a considerable amount of operational expenditure if they can embrace a similar model.

But that’s not how carriers run their networks today. “It’s not just a question of adapting to new technology - it requires a change in culture and skill sets and that just doesn’t happen overnight”, says Elby.

The ONF’s Pitt accepts that the carrier world is cautious, but warns against ignoring the march of SDN: “Carriers must understand that they will have to become software companies – that software will be the key to their differentiation in the future.” That in the end will gate the timing of wide-scale SDN deployment. “We’ll close the technology gap much quicker than we will be able to move the engineering and operations workforce over to the new paradigm”, says Elby.

What’s the difference between software defined networking and the OpenFlow Protocol?
SDN is a concept that separates the intelligent control of a network from the transport function and subsequently hands that control to a programmable computer. OpenFlow is a computer language that allows the controller to talk back to the network.

How big is the potential market?
According to the information provider IDC, sales of SDN gear as well as software and services will surge from $200 million in 2013 to $2 billion in 2016. Some analysts put the number closer to $10 billion over a similar time frame.

Do carriers really have that much budget to throw at SDN?
No one is building greenfield networks anymore so any move to SDN will necessarily have to be an evolutionary process. But everyone is looking to move as quickly as possible to a lower cost structure: the quicker SDN architectures help better utilise the network, the faster carriers can grow. That’s a compelling argument for most finance directors.

How long will it take before SDN comes of age?
Don’t expect to see carriers try any meaningful deployment in their wide area networks much before 2015 – SDN is currently undergoing rigorous testing among several prominent carriers including NTT and Verizon. It’ll take about 12-18 months to iron out any glitches and at least a further 12-18 months on top of that to scale up for a launch.

What’s the killer App?
No one really knows yet. There’s a lot of low hanging fruit to pick around network optimisation, traffic flow and security. But like the iPad, SDN may never foster a killer App – just scores of smaller apps that carriers never knew they couldn’t do without until they were launched on the market.

Who are the companies to watch?
The SDN arena is a hotbed of M&A activity at the moment. Nicira, whose chief technology officer helped devise OpenFlow, recently fell to VMWare for $1.2 billion while Oracle snapped up San Jose-based Xsigo Systems for an undisclosed sum. Other emerging start-ups include Big Switch and Arista Networks. Don’t be surprised to see IBM, Cisco or even Juniper step out along the acquisition trail.

So is this the end for Cisco?
Sales of routers will probably fall as carriers learn to optimise their networks and simpler switches will undoubtedly come a lot cheaper. But remember, SDN is all about mining the intelligence off the network and currently, that sits on Cisco gear. No one knows better than Cisco how to harvest that intelligence.

Can SDN live up to all the hype?
You bet. Google has already migrated one of the largest networks in the world over to an SDN format and expects to see cost savings as a result. If that’s not enough to tempt carriers, the prospect of developing new services that can attract big premiums surely will.