Software Defined Networking? It’s Just Correcting An Error

stu bailey infoblox

The industry needs software defined networking, to correct its over-dependence on specific hardware, says Stu Bailey

Software defined networking (SDN) is generating a lot of excitement in data centres and other networks, but it is really a simple market correction – overcoming the industry’s fixation on proprietary hardware boxes for generic functions.

stu bailey infoblox 2That’s the view of Stu Bailey, CTO and founder of a company which found itself at the heart of SDN almost by accident – Infoblox.

Understanding software defined networking

“We are in a very good position to help the market understand what SDN is,” Bailey told TechWeekEurope. “We are not disrupted by it – but we are not betting on it like some of the startups.”

Bailey started Infoblox in 1999, with a mission to make networks simpler. This started with appliances to handle repetitive network tasks such as managing DNS, DHCP and the like, calling itself the “iTunes of IP” – and TechWeekEurope has referred to it by the somewhat dismissive handle of “network housekeeper.”

By the time the company had its IPO in April, its mission was broadened to become what Bailey calls “the leader in control plane software for networking”.

When the software-defined networking story “broke” about 18 months ago, Infoblox found itself in a really good place – the incumbent vendor in the network control plane space, which SDN was about to turn into a crucial area for network vendors. Infoblox took a vocal role in the Open Networking Foundation, where the OpenFlow standard is owned.

SDN corrects hardware defined networking

So, what does he say about SDN?  “Before you understand SDN, you have to ask what is hardware defined networking,” says Bailey. He believes that SDN is a technology story, in which the old approach where networks were defined by hardware will inevitably be displaced.

“The striking thing about hardware defined networking, is that it is an entire huge industry organised along box function lines,” he says. “Vendors, analysts, and even the way internal IT teams at user companies are organised. The entire business is structured around firewalls, load balancers, WAN optimisers, and so on.”

The industry needs to change, he says, because of a fundamental falsehood in the way the industry works now. While vendors are busily selling firewalls and switches as if they are different objects, they are in fact, pretty much the same. “They are fast, they are cheap and they are Ethernet,” he says. “The differentiator is in the software.”

While boxes are becoming more commoditised, the users aren’t getting the economic value that buyers of other equipment – say servers – are getting as hardware simplifies.

“This is really a market correction that has been building for several years,” he says. “On the other side of this correction, organisations that need scalable network infrastructure will have a different method of deploying it. They will have ultra-commoditised hardware and rich software.” The value in networking, he says, will shift to the control plane – software not hardware.

The change is similar to the death of standalone word processors, killed by the PC, or the mainframe, whose progress was halted by general purpose computers from minis to blade servers. The big marker, for him was when virtualisation player VMware, paid $1.26 billion for SDN start-up Nicira in July 2012.

“It is an industrial value shift,” he says, “shifting from box function to cheap hardware and value added software is a big deal for the landscape of vendors. It is a bigger trend than we saw in virtualisation. It is more fundamental – because servers were already general purpose before being virtualised. The virtualisation trend was about consolidation; SDN is being driven by a fundamental shift in economics.”

What about the big boys?

That begs a big question: What will this do to the network incumbents?

“The landscape of vendors in hardware-defined network is going to be radically disrupted,” says Bailey. “There will be two classes of vendors: those that are focused on the software, who can solve the complexity problems that are a barrier to scale, and those focused on low-margin commodity hardware to support that software.”

Bailey won’t predict what individual firms will do, but he does believe that “Those who buy network infrastructure will have a much better set of products that attack the problems of complexity and cost, and this will open up opportunities for emerging vendors. I don’t think we have seen this kind of movement in a major space since the 1980s.”

I ask where Cisco will go, and he won’t be drawn. It seems to us that the incumbent vendors can’t make that decision. Cisco’s CTO, Padmasree Warrior maintains Cisco can weather the SDN storm by offering a superset of the standard on its hardware, maintaining the special role of Cisco switches, and we can’t see how it can move. “Cisco is complacent,” he says, “that is exactly what IBM said in the 1980s.”

We push him. Should Cisco change to become a commodity hardware vendor? That would throw away one of the industry’s most enviable margins – albeit possibly just before someone else takes that margin away from it, but why throw it away pre-emptively?  If it moves to be software-only, it loses the hardware completely. If it splits into two, we suggest, the sum of the parts is inevitably less then the whole, as that is the fundamental correction that Bailey is proposing.

Maybe, but he won’t predict which direction the network giant will move. “If they become the standard for commoditised cheap hardware, and truly make an ecosystem for third parties, that would be fundamentally disruptive to the way their channel is structured. Conversely if they really focus on software, they have to standardise on a hardware platform that is not theirs. This is all fundamentally disruptive to their worldview.”

This will affect everyone

But Cisco’s rivals face the same problems: “This is equally disruptive to Juniper – it is really about the hardware defined network, business.”

Infoblox is small potatoes compared to these giants, but it’s been software-defined since the start, even though it used to sell only appliances. The appliances are still there, but hardware is only incidental, says Bailey – Infoblox only charges ten percent more when the software is delivered on hardware, compared with just getting the software delivered virtually. And all of Infoblox’s new products will be software-only.

“We are well positioned to help in the transition,” says Bailey. “We understand this is a major shift in how people buy and think about networks.”

How quickly will the change happen? Bailey admits he doesn’t know, but says the change will be “dramatic”, and “today is when you should be planning a strategy to understand how your market is going to react to this market correction.”

“When the right pieces fall into place, it will be very fast and transformative  in the same way that the Web was, client server was, and the PC was: “When these things really  started to happen, it changed the landscape of vendors.” And it’s already been starting to happen for a while, he says, as vendors have begun to take advantage of cheap hardware, and adding value in the software.

But despite the apocalyptic tone, this isn’t all bad news: “Efficient markets are healthy. The more the market can be prepared for this shift, the more everybody wins. Vendors get value, consumers get value, and the market is healthier.”

Yikes – better do our Cisco quiz, while the network giant is still around!