Carriers are starting to adopt software-defined networking (SDN) in earnest, having moved out of trial phase and into serious real-life commitment.
But as with any deployment of a new idea, there is an adoption spectrum that ranges from enthusiastic first-movers to conservatives with an entrenched “wait and see” policy. The danger here seems obvious – that a two-speed migration might isolate the slower carriers and prompt the faster movers to do business primarily – or even exclusively – with other fast-moving disciples of software-driven connectivity.
To look at the communications and data ecosystem in its widest sense is, of course, already to see a distinct schism. Within the world of the data centre, the big Web 2.0 players like Facebook and Google have for some time been setting the pace in software-defined activity. Within the latest generation of data centres, technologies like Open Compute, white box switches, SDN and network functions virtualisation (NFV) are common currency. There is no talk of first movers here. The early moving has already been done.
Among carriers, however, progress has been a lot slower. A number are well ahead with NFV deployment at Layer 4 or above, manifesting in the form of virtualised broadband remote access servers (BRAS) and firewalls. But SDN is a trickier prospect for them than it is for data centre operators. Wide-area networks (WANs) are complex things, covering in many cases vast geographical distances. Not all carriers share the same view of the ultimate framework, information model and protocols that make up a WAN, meaning there is no foolproof and obvious SDN template that works for everybody. Anyone wanting to be known as a first mover will understand that they are taking the risk of building where the foundations have arguably not yet solidified.
“Actually we think it might be best for a carrier to be a fast follower instead of a first mover,” says Mike Capuano, VP of corporate marketing with WAN equipment maker Infinera. “We believe it is prudent for carriers to push their vendors hard for SDN and to take an approach where they collaborate and experiment with their vendors.”
He says Infinera has worked on developing SDN with organisations ranging from the Energy Sciences Network to Telefónica, prototyping multi-layer optimisation with the former and network-as-a-service with the latter.
“It has to be said though that there are a few first movers, like Pacnet, that have gotten something out quickly, learned from that deployment and are on to the next release,” he concedes. “There is no doubt Pacnet has received a lot of visibility for this.”
Pacnet’s reputation as an SDN pace setter is certainly merited, if for no other reason than the decisiveness with which its strategy was carried out. Jon Vestal, VP of product architecture with Pacnet, explains that once the idea of the Pacnet Enabled Network (PEN) – a service platform offering scalable bandwidth – had been hit upon, the Asian wholesaler was determined not to waste time: “We started in May 2013 and went from idea to beta to the finished thing in a few months, which is good for a carrier,” he says.
He insists, though, that the other plank of Pacnet’s SDN strategy was keeping customers in the loop from the outset: “What was also different, apart from moving so quickly, is that we were talking to our customers during the production process, through focus groups and other such means,” he explains. “This is definitely not something we’ve done often in the past. We’ve now got the product that customers want, with its own open APIs and open pricing too. The customer can choose it the way they want, and make their own decisions on price performance. They can even build their own policies. We’ve innovated technically and commercially.”
Vestal offers reassurance that Pacnet is not about to embark on doing business only with other SDN-friendly parties, reasoning that SDN has some way to go before this would even be possible, let alone desirable.
“I was at a conference a couple of months ago and someone was asking ‘is SDN an all-or-nothing game?’” he recalls. ”The answer I’d give is ‘No’. SDN offers a new way to do things, but it has some limitations. Not all aspects of a network will be SDN-capable for some time, and I think there will always be a hybrid element.”
He says that when choosing a partner to strike a network-to-network interconnection agreement (NNI) with in a new territory, Pacnet might well prefer another SDN-enabled network, but has to be primarily mindful of the needs of the customers on whose behalf it is seeking the NNI in the first place: “An enterprise might be able to afford a rigid approach, but as a carrier we need to work with all of our customer base,” he adds.
Looking for a framework
As carriers move to SDN, they have the option of adhering to the Open Networking Foundation’s (ONF) OpenFlow protocol, either in its entirety or in part. Some are simply opting to deploy a form of software-enabled network automation and are getting benefits from that without getting anywhere close to what some would call full SDN implementation. Few carriers are strict OpenFlow adherents, in all its complexity and rich programmability, at this stage.
This doesn’t seem to faze Dan Pitt, ONF executive director.
“Standards are crucial, but not as important as adoption,” he argues. “We’re advancing rapidly with OpenFlow, in the area of optical transport and also in wireless too. It’s important to expand the application of SDN to more environments. We’re working on the security side too, and we’ll be making more announcements on that later in the year. Interoperability is another focus for us. We held an event simultaneously in the US and Beijing to demonstrate cross-site possibilities.”
He says the ONF is also working with other SDN initiatives, like OpenStack and OpenDaylight. In the absence of complete agreement on open standards for SDN, Pitt believes matters are “trending in the right direction”.
“There are many parties participating with what we are doing, and great variety between them,” he claims. “There is of course a spectrum of visions out there, with some ‘open’ solutions still quite proprietary. It’s not necessarily a bad sign if operators are all doing it in different ways. After all, they all have different customers and face different pressures. I’m generally pleased to see what they are doing with SDN. In Asia, for example, it’s amazing to see how fast they are moving, less encumbered by legacy investment as they are in many cases.”
Choosing tactics
Chris Emmons, director of network planning with Verizon, thinks most carriers are still at the stage of making their mind up on exactly what SDN will mean to them.
“Everyone is at the stage of figuring out which way the market will go,” he says. “You could call it ‘testing the threat’. If you ask where we are at, I’d say we’re selectively pioneering on SDN. In some areas of development we’re well ahead and on others we’re adopting a more wait-and-see approach. It’s at a stage where nobody can afford to be full-on bleeding edge any more than anybody can afford to lag totally and do nothing.”
Antonio Durán Domínguez, wholesale product manager with Telefónica Global Solutions, agrees that many carriers are still assessing the evolutionary path of SDN.
“There are many positive aspects to SDN, but equally there are hurdles to overcome,” he suggests. “Perhaps it is most appropriate to consider 2014 a ‘proof of concept’ year, with many of the industry players running labs and associated PoCs. At this stage however, it is clear that there are no universal standards which are essential for the wholesale segment to underpin interoperability.”
Wholesalers, as opposed to their retail customers, may have greater motivation to move sooner rather than later on SDN. The efficiencies that SDN promises are probably not enough at the moment to justify a small regional service provider in replacing infrastructure they already have. But for a wholesaler whose business is based on tiny margins and arbitrage, the dividend of SDN might be important, or even critical. It could bestow significant competitive advantage for those that have it, giving them control over their volumes of traffic and allowing them to more effectively face OTT challengers.
All telcos, wholesale or retail, will have an SDN strategy of sorts, but where they are on executing it will depend on the extent of their top-line motivation, suggests Kelly Herrell, VP and GM of software networking with vendor Brocade.
“It depends on whether they are just selling connectivity or applications as well,” he suggests. “Cloud providers are ahead on SDN because their motivation is higher. They’ve already had to figure out the automation of their network’s software. The service provider, on the other hand, owns vast networks where software could potentially be used. But where do they start?”
One point of entry, he says, could be creating a ‘subscriber cloud’: “The ways service providers are currently managing subscribers is like they are using stone tools,” he explains. “New technologies can make you more agile here, and that might affect your desire for change. Should they be worrying about where they are on the spectrum? I think their focus should be more on making it work for themselves.”
The worry around SDN implementation among carriers may simply be a response to a tsunami of industry hype that they should already be familiar with, says Matthew Finnie, CTO of Interoute. “The problem is that people suddenly think it’s about software-defined everything,” Finnie complains. “There’s a lot of agendas out there, some of which are a bag of nonsense. I see SDN more as a natural evolution of something that we are already doing. A bigger issue by far is the convergence of computing and networking.”
Somewhere on the curve
At this stage of SDN’s evolutionary curve, many carriers will be looking for a model that manages both physical and virtualised aspects of the network. We may be moving towards a virtual future where physical distance between computing power and application usage no longer matters, but at present certain applications still need highly proximate computer power, and virtualisation may not make sense for them today.
There is also the matter of allowing existing investment to live out its lifespan: “You have to be able to automate existing infrastructure,” agrees Sunil Khandekar, CEO of SDN-focussed vendor Nuage Networks, a spin-off from Alcatel-Lucent that launched in early 2012. “Cisco’s SDN model only works on new physical infrastructure. But if the customer has invested in building a data centre, they won’t want to rip and replace just to see the benefits of SDN. What’s the object of SDN? It’s to bridge the gap between network and application, bringing about tremendous agility. You’re restricted if you can only deploy on new infrastructure.”
Any successful SDN strategy should be determined not by a clash of technology choices, but by the services to be deployed and marketed over the network. It’s here that SDN qualities like agility and flexibility matter, argues Phil Braden, SVP technology and applications with PCCW Global.
“We see it as agile in as much as we are able to deliver new services with it without redeploying the network and without affecting existing services,” he explains. “And it’s dynamic because we are able to deliver services on demand, in response to customer or application activity, or in response to changing network conditions. It’s manageable in the sense that we are able to manage the network effectively and control the flow of data and the load across the network. And it’s assured because we’re able to manage and assure QoS of traffic.”
He insists that PCCW’s choice of global partners would never come exclusively from a pot of pioneers in adopting a specific technology, but rather those who are able to match the standards of services it strives to offer based around the above qualities.
SDN may not then be developing into an exclusive members-only club, but it is moving fast. It should also be judged as a long-term movement that is here to stay and not a quick fix to be rushed into.
Carriers should at least be realising by this stage that they cannot just carry on in the old way forever.
