As executive director of the Open Networking Foundation (ONF), Dan Pitt has remained a key figure in software-defined networking (SDN) even as the concept’s scope has widened. Pitt was a featured speaker at the recent SDN & OpenFlow World Congress, in Dusseldorf, Germany, where he caught us up on what’s happening with OpenFlow and the ONF, and explained why SDN‘s open source movement has been a little disappointing.
How would you characterize the state of OpenFlow?
I think it’s finally going to be mainstream. It’s going to be the default southbound protocol. [Vendors’ attitude will be:] “I’m going to take that; there’s no compelling reason why I can’t,” as opposed to, “I’m going to do everything I can to avoid it,” which — we’ve seen some of that behavior.
Why couldn’t you say the same last year?
Partly, we haven’t had the hardware support. We haven’t had the conformance [testing] for [OpenFlow] 1.3 — this week, we’re announcing a conformance pilot program with nine companies in it. And the progress we’ve made with the chip vendors — the next 12 months are going to be big for that.
I talked to a major incumbent vendor that has a software line; it’s kind of an overlay approach. I said, “What are you doing with OpenFlow?” “Oh, we use OpenFlow. It’s not like we sell it. it does what we need it to do so we can sell products.” They’re using it instead of contorting themselves to avoid it. It gives them the control that they want.
Prodip [Sen, of HP] made the comment that the operators don’t have to adopt open source software themselves at all. Their vendors will adopt it and make products out of it. My concern is — yes, it is more developmentally efficient for a bunch of people people to cooperate on code for certain purposes where they don’t have to differentiate, and to deliver that code in proprietary products. So it’s just a shortcut to development.
Whereas, what I was hoping for — what a lot of people were hoping for — was that some of these big efforts would result in free open-source products. And maybe you get a Red Hat-like model — somebody will offer support and service — but if you want to, you can just take it, as opposed to having code sucked out of [an open source project] and put into proprietary products.
That’s the trend that I am less happy with. There’s a benefit to it, but it leaves me unsatisfied with what I think is the whole promise of these large open-source projects.
I get the feeling you’re still optimistic about open source, though.
There are going to be some really good pieces that come out, and they might not be the things you expected — like Teflon from the space program.
That’s part of the fun. Some open source projects will shake out in ways you don’t expect.
People complain about the chaos. “It’s too immature.” But to me, chaos is a sign of all kinds of ideas being tried. And yeah, I’m as eager as the next guy to say, “Let’s get some solutions we can deploy,” but I find this part exciting. You have to enjoy the journey, here. Don’t wait ’til you get to the destination. You’ll never know when you’re there.
There are so many working groups within the ONF now. Do you worry that it’s getting too complicated?
We did a reorg to make it easier for operators to participate with all these working groups, to get software more into what we do, and to scale. We took working group discussions and we put them into so-called areas, just like the IETF, and we pushed a lot of decision-making down to area directors: how to organize them, group charters — get them producing stuff, get a schedule.
We also formed a bunch of councils — technical, executive, CTO, and operator — and the executive council includes people from some member companies that can help me run this organization. I don’t have the usual complement of vice presidents, a chief operating officer, a CMO. Our “vice presidents” really come from industry. So, companies represented there include China Mobile, Equinix, Prodip from HP, and Justin Dustzadeh from Huawei.
So, that’ll give me some help and offload the board a little bit. It’s just to run faster and do better, and push a lot of decision-making further down in the organization. It’s the only way to scale.
These days, where do you see the ONF fitting in? Back when, you told me the ONF’s goal was to push SDN out to the world. But now, there are so many other groups doing that.
Yeah, but in various ways. Some are just doing their own thing. Others we’re working with. We’re working with the OIF. We have some cooperation things going on with OpenDaylight — TTPs [table type patterns] and northbound interfaces — so we’re working jointly on those things. We’re working with the UCI Forum for unified communications, also on northbound interface stuff.
We’ve got the expertise below the interface between us, and they have the expertise up above. So we’ll meet in the middle and make sure that meeting is successful. We’re happy to lend our expertise. We want SDN to be exploited by all the software services and models that run above.
And we’re still doing the infrastructure stuff that has to be done. No one else is going to do a forwarding plane, and we want that to be a no-brainer.
SDxCentral is in Dusseldorf, Germany for the 2014 SDN & OpenFlow World Congress. Check out all our headlines from the show here: SDN & OpenFlow World Congress: The SDxCentral Report