10:15 AM
Mike Fratto
Mike Fratto
Connect Directly
Repost This

Prediction: OpenFlow Is Dead by 2014; SDN Reborn in Network Management

OpenFlow and software-defined networking were on the agenda at Interop yesterday. But will they stand the test of time, or will SDN eventually become part of network management?

Jim Metzler and I co-chaired an afternoon workshop at Interop yesterday: "How Will Software-Defined Networks and OpenFlow Impact Enterprise Networks?." It introduced OpenFlow and software-defined networking (SDN) and highlighted some use cases.

Despite that, in two years I think the OpenFlow bubble will burst, and SDN will sink into the Trough Disillusionment on Gartner's Hype Cycle. It will be reborn as part of network management.

At the end of the day, someone in the audience asked a question along the lines of why Guido Appenzeller, CEO of Big Switch Networks, and Bruce Davie, chief service provider architect at Nicira, thought organizations would adopt OpenFlow in 12 to 18 months. They gave the same answer they have been giving all day: The adoption will be slow and will be used only where it fits, etc. I can't argue with that.

Or can I?

It was the end of the day. Maybe I was grumpy. Perhaps I wanted a cookie. Maybe I just wanted to make a statement. So I probably said the most words I had said all day (or something like this): "I think you two are being too optimistic. The cards are stacked against your two companies, OpenFlow and SDN. It took VMware about seven years to reach critical mass. IT shops are conservative. They don't want to monkey with the network running mission-critical services. They don't want to monkey with part of the network running mission-critical services, for fear something will break. That's a huge hurdle.

"Nor will the big networking vendors go away silently," I continued. "They are going to muddy the water, calling their APIs and SDK--none of which come close to 100% feature coverage--SDN, and even if they did have 10% coverage, it still wouldn't allow you to control flow paths. And let's not forget all the application vendors that had to be dragged kicking and screaming into server virtualization, thinking--wrongly--that their applications need eight cores and 32Gbytes of RAM when, in reality, they only needed 10% of that.

"They aren't going to want some new networking technology interconnecting their stuff, nor will any equipment vendors that have predefined hardware compatibility lists. Like storage vendors who, once you go off the HCL, point out, your kit doesn't comply with their best practice and you get no support until you bring it into line. Those factors have nothing to do with your products or technology--all of which may be great, creative and useful--but factors beyond your control are going to be a significant drag on adoption."

Davie made a funny but pointed statement, saying we don't want IT to give up its reliable and robust network. We want IT to keep it, and that way our stuff runs even better. While his answer is somewhat tongue-in-cheek, his point is that IT still wants fast, reliable hardware--Nicira just wants to manage it better.

I talked with Appenzeller a bit after the session, and he suggested I look at all the vendors lined up against Cisco, not to mention speak with potential customers wanting an alternative to the networking giant. I'm sure Appenzeller has had those conversations--you can hardly swing a Catalyst 6500 at Interop without hitting disgruntled Cisco customers. But are those customers disgruntled enough to either switch or even augment what they already have with Cisco?

No, I don't think so. Sure, big Internet companies like Google, cloud providers, online stores and other companies that make high demands on their networks and have the staff to engineer them can make a go of OpenFlow and SDN. But in my wholly unscientific opinion, I bet most companies aren't going to seriously consider either OpenFlow or SDN for some years--a long enough time for the VC money to dry up and larger vendors to pick up the carcasses along the Internet superhighway.

Don't get me wrong: The technology these SDN startups are developing is innovative, creative and challenges the status quo. They seem to be offering useful alternatives in how IT designs and runs their networks today. William Koss sums up what SDN startups are doing in "Missing the Point on Software Defined Networking (SDN)": "We should be intrigued that people are asking: 'Is there a better way?' ... This is also the reason why I think the architects, the revolutionaries, the entrepreneurs, the leaders of the next 20 years of networking are not working at the incumbents."

Of course, many of the problems that the SDN vendors state--VM mobility, the limited range of VLAN IDs, the inability to move L2/L3 networking among data centers and the inflexibility of current networking command and control--are problems faced only by cloud providers and a handful of large, large companies with big, global data centers. In other words: a rather small number of customers.

Many of the SDN vendors will be acquired by the incumbents, and perhaps the technology will make its way into shipping products. Until then, IT will continue to do whatever it is it does inefficiently.

Comment  | 
Print  | 
More Insights
Newest First  |  Oldest First  |  Threaded View
User Rank: Apprentice
5/21/2013 | 2:39:06 PM
re: Prediction: OpenFlow Is Dead by 2014; SDN Reborn in Network Management
Do you know COBOL, right? It's been used for years, because "we don't want IT to give up its reliable and robust banking systems". But nowadays Java, Ruby, .Net and other technologies are a much better replacement. Besides, must of them can inter-operate with COBOL if necessary.
User Rank: Apprentice
5/9/2012 | 6:52:19 AM
re: Prediction: OpenFlow Is Dead by 2014; SDN Reborn in Network Management
Such a delightful bit of curmudgeoney-goodness. You usually have to head on down to the "Old Networker's Bar & Grill" and hunker down in the smoky backroom with the grizzled IT writers in their dusty fedoras and overcoats over boilermakers of Blatz and Old Crow to dredge up such a refreshing perspective as 'networks will never change...'. It's probably up there on the wall, along with "Virtualization on X86 will never happen..." and Ethernet isn't suitable for storage traffic". If you look around a little, you're bound to notice that these old ideas are being shattered every day; in part because good technology usually finds a way, and OpenFlow is good tech.

Fortunately, there are many of us thay don't buy the premise that network admins are - above all - scared of anything that might bring a network down. (Of course, this also assumes that OpenFlow is somehow less reliable than existing network control protocols). Though FUD still tends to rule in the short-term, in the longer term a strong alternative technology like Software Defined Networking; that has the potential to improve network efficiency, lower costs and reduce dependence on proprietary solutions, may eventually gain a foothold.

So it ultimately becomes a question of how to define the success or failure of a concept like SDN? You imply that all it takes to run OpenFlow off the rails is for vendors to bifurcate the published standards with their own versions; making the open standard irrelevent. This has happened in the past, after all vendors will be vendors and noone likes to give up marketshare. But oddly enough, rather than fighting an upstart technology like OpenFlow we're starting to see vendors like IBM and HP adopting the standard without a lot of fanfare; and I predict that there will be more. Many more. Apparently it's not that hard to embrace OpenFlow, so why not?

Sure, there will be a lot of churn and burn in the networking industry over SDN, but is that necessarily a bad thing? Competition drives innovation. And the networking industry, which has become amazingly stagnant compared to other segments of IT, could stand to use a bit of a shake-up.

That being said, I don't really want to sound like I'm buying into the hype-cycle either. Are SDNs capable of handling everything that conventional networks do? Not yet. Will OpenFlow mean the end of homogenous networking hardware? No. Does the ONF have the power to enforce the OpenFlow standard. No.

On the other hand, can SDNs scale to fit high-performance, hi-reliability networking applications? Yes. Can OpenFlow give admins more options when it comes to choosing non-standard yet best-of-breed hardware where it's needed most? Yes. Does OpenFlow offer a much needed update to the creation and management of modern networks? Yes. Will this take time? Most assuredly yes. But expecting such a major reboot of the networking industry to make a major impact in less than 5 years isn't close to being reasonable.

Of course this brings up the inevitable Catch-22 regarding any kind of change in IT. If an idea is less than 2 years old it's "untested", less than 4 years old is "immature", and over 6 years is already on it's way out. Apparently, there's no winning unless there's immediate and massive adoption, but by that standard a technology like server virtualization, or 10Gb Ethernet that taking years to gain marketshare is technically a failure.

Let's not toss the baby out with the bathwater. OpenFlow is the most interesting thing that's happened to networking in AGES, and rather than dismissing it offhand I think that I'll give this particular upstart a chance to succeed. Fly, little bird, fly...

Steven Hill - NWC Mod
More Blogs from Commentary
Infrastructure Challenge: Build Your Community
Network Computing provides the platform; help us make it your community.
Edge Devices Are The Brains Of The Network
In any type of network, the edge is where all the action takes place. Think of the edge as the brains of the network, while the core is just the dumb muscle.
SDN: Waiting For The Trickle-Down Effect
Like server virtualization and 10 Gigabit Ethernet, SDN will eventually become a technology that small and midsized enterprises can use. But it's going to require some new packaging.
IT Certification Exam Success In 4 Steps
There are no shortcuts to obtaining passing scores, but focusing on key fundamentals of proper study and preparation will help you master the art of certification.
VMware's VSAN Benchmarks: Under The Hood
VMware touted flashy numbers in recently published performance benchmarks, but a closer examination of its VSAN testing shows why customers shouldn't expect the same results with their real-world applications.
Hot Topics
IT Certification Exam Success In 4 Steps
Amy Arnold, CCNP/DP/Voice,  4/22/2014
Edge Devices Are The Brains Of The Network
Orhan Ergun, Network Architect,  4/23/2014
Heartbleed Flaw Exploited In VPN Attack
Mathew J. Schwartz 4/21/2014
White Papers
Register for Network Computing Newsletters
Current Issue
Twitter Feed