Blog Post

Is OpenFlow an answer looking for a problem?

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!
Lane Patterson Equinix Kenneth Duda Arista Networks Structure 2012
(L to R) Lane Patterson, CTO, Equinix; Kenneth Duda, Founder, CTO and SVP, Software Engineering, Arista Networks
(c)2012 Pinar Ozger [email protected]

OpenFlow and software-defined networking may be hot topics in the infrastructure industry right now — but not everyone believes they are up to scratch.

At a panel discussing internet bypass and new data center fabrics at GigaOM’s Structure 2012 in San Francisco on Thursday, two senior industry figures said that the standard was still too immature to be useable.

Kenneth Duda, CTO and founder of cloud networking provider Arista Networks, launched a stinging criticism of OpenFlow — which he said was unable to do many of the jobs currently being ascribed to it.

“I think there’s a lot of attention placed in this space, and a lot of customer aspirations in this space that makes a lot of sense,” he said. “It seems to me that the promises have gotten a little bit of a head of delivery in some cases.”

He explained his thinking by suggesting a potential use case.

“There’s one application where, if you’re a government agency with a three letter acronym and you come in possession of 10 terabits of snooped network traffic, which you’re not allowed to store because it would violate federal wiretapping law, what you need to do is distribute this traffic over a bank of servers for analysis as quickly as possible,” he said.

“But then send http this way, send voice that way, send video that way, send email this way, subdivide it further by geography, by anything you know about the parties communicating and so on, and do all this in real time to balance load.”

“It turns out OpenFlow 1.0 is a very good fit for that problem: I haven’t found any other problems for which I believe OpenFlow 1.0 is a good fit.”

The comments were echoed — though not as strongly worded — by Lane Patterson, CTO of Equinix (s EQIX), who said that while OpenFlow looks promising, it still has a long way to go.

“We’re not quite sold on the maturity yet, so it’s still something we’re playing with in the labs,” he said.

“What OpenFlow represents for an end user or a service provider like Equinix… as we’re moving into this market and services creation space, and facilitating buying and selling through automated, cloudified means, we’re really looking for a vendor-independent way that’ll become a standard. Everything we’re doing today, we’re adapting to each vendor’s specific platform.”

Check out the rest of our Structure 2012 coverage, as well as the live stream, here.

Watch live streaming video from gigaomstructure at

2 Responses to “Is OpenFlow an answer looking for a problem?”

  1. Casimer DeCusatis

    While OpenFlow is admittedly not fully mature yet, the ONF is gradually releasing a series of use cases for OpenFlow based on client feedback. These should help clarify the types of problems OpenFlow can solve. I don’t think Google, Verizon, Deutche Telecom, and others would be investing their time in the ONF if they didn’t have use cases in mind. It also sounds like Equinex is supporting OpenFlow in this article, by saying that they’re looking for a standardized way to manage multi-vendor environments. For more on standards based solutions and SDN/OpenFlow, see the ODIN materials at

  2. Reblogged this on Virtualized Geek and commented:
    Openflow was a secondary topic for me in this video. The real meat was Arista talking about Internet Bypass. The idea that they can or will create a broker service that will be “Cloudified” or orchaestrated to set up peer to peer BGP connections between the provider and customer is a serious piece of innovation. I would have loved to hear someone from a cloud data center provider comment on this technology.
    It’s one thing to talk about orchestrating servers, hypervisors and datacenter networking but I have to imagine providers are loath to let cloud tools automate the creation of BGP routing at the edge of their network. This is where you can break the Internet but I see the appeal. I would have liked to be in this Q & A.