Networking

10:06 AM
Dan Pitt
Dan Pitt
Commentary
50%
50%

SDN: What's In Store For 2014

Software-defined networking has been THE topic in the networking industry this year. As 2013 comes to a close, let's take stock of its progress and consider where it's headed.

Many will agree that SDN standards such as OpenFlow have come of age in 2013. By the end of 2014, I believe that all switches will have OpenFlow client support. Demand for SDN is building, putting pressure on networking vendors to accelerate their support for OpenFlow. Soon it will cost more not to have an OpenFlow client in a switch than to have one, so it will become a standard feature.

This leads to my other predictions on the changes that we'll see in SDN next year:

Interoperability will actually happen. Numerous SDN controllers and OpenFlow-enabled switches and routers have already come to market, including the first generally available SDN controller (from NEC) supporting OpenFlow 1.3. As the OpenFlow substrate continues to build out, the focus in 2014 will be on interoperability.

To that end, the Open Networking Foundation (ONF) this year kicked off its OpenFlow Conformance Testing Program, enabling vendors to demonstrate conformance with the OpenFlow specification. Currently there are three conformance testing labs (two in the US and one in China), and in October an NEC product received the first Certificate of Conformance. In the coming year, we'll see more testing done as OpenFlow conformance becomes a marketing necessity. More conformant products will be hitting the market and showcased in proofs of concept.

[Need to learn the basics of SDN? Check out our slideshow, "7 Essentials of Software-Defined Networking."]

Telcos will help drive SDN adoption. Telecommunication service providers made network functions virtualization (NFV) and SDN hot issues in 2013, and they will be a major purchasing force in the SDN market in 2014. For example, AT&T announced that its Domain 2.0 initiative will use SDN and NFV technologies and architectural approaches to simplify and scale its network. ONF will help these operators by enabling NFV with SDN and OpenFlow.

The vendor landscape will change. This year AT&T issued the first major RFQ specifically calling for SDN support, including demands for the separation of hardware and software. The carrier went so far as to note that delivering SDN and NFV "will require some of AT&T's current providers, but also will require some new providers with different skills and capabilities." That smells like opportunity -- at least for some.

New SDN-focused vendors will be key to the delivery of separate, interoperable hardware and software components. Just look at all the new startup companies, including not just software but even chips. ONF will continue to help build an ecosystem that encourages both innovation by the incumbents and the formation of new companies focused on network transformation.

 

 

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
Susan Fogarty
50%
50%
Susan Fogarty,
User Rank: Strategist
12/13/2013 | 9:51:04 AM
Re: SDN == OpenFlow?
ACM, that's a sobering observation. Do you think IT professionals have actually been buying into the hype and think they can easily implement SDN, and they'll be surprised when it's difficult? It seems awfully complex to me and just because Google's using it doesn't mean your average data center can.
Drew Conry-Murray
50%
50%
Drew Conry-Murray,
User Rank: Strategist
12/11/2013 | 1:38:20 PM
Re: SDN == OpenFlow?
2013 was the year of SDN promises. 2014 will be the year of dissapointment as companies actually start with deployments and find out how much work still has to happen, both from vendors and inside customer data centers.

 
@mbushong
50%
50%
@mbushong,
User Rank: Moderator
12/11/2013 | 11:19:15 AM
Re: SDN == OpenFlow?
Sure, deployments will serve as references for other deployments. The question is whether you measure penetration by number of deployments, by size of deployment, or some combination. SDN is most easily deployed in green field opportunities, because you can build from scratch and reduce the dependency on legacy interoperability. This opens up more potential solutions. 

These opportunities will, IMO, tend to be smaller (start small, right?). And the types of applications that are seeing traction (think: Tap applications, among others) will tend to be simpler.

When people bridge from small, starter deployment to something larger, everyone (both the customer and the vendor) will learn something (a good thing!). That learning will be on the operational side (also just IMO). Those operational things need to be learned and addressed before deployments get bigger.

I don't mean to suggest we won't see success in 2014. I just don't think it will be of the massive or heterogenous varieties. There is still a lot to be gained for companies, and the seed deployments will be very important. So we should still exit the year with Hope.

-Mike
Susan Fogarty
50%
50%
Susan Fogarty,
User Rank: Strategist
12/11/2013 | 10:52:05 AM
Re: SDN == OpenFlow?
Wow, @mbushong, you are quick on the uptake! Don't you think if there are some deployments, even if they are single vendor and very controlled, that will give some confidence to other potential customers? Companies can experience some benefits without software-defining their entire environment at once, and they are going to have to take it one step at a time.
@mbushong
50%
50%
@mbushong,
User Rank: Moderator
12/10/2013 | 8:40:05 PM
Re: SDN == OpenFlow?
I think you will see deployments. I just don't think they will be of the massive heterogeneous type. I think you will see vendors have success with their individual solutions, and you will see a handful of truly heterogeneous success stories (I would guess R&E is where those show up first).
Susan Fogarty
50%
50%
Susan Fogarty,
User Rank: Strategist
12/10/2013 | 8:29:13 PM
Re: SDN == OpenFlow?
@mbushong, I agree that there is no "winning" approach to SDN as of yet -- it's still anybody's game. That said, I am hoping 2014 sees a little more progress than you predict. It definitely takes a great deal of expertise and it not for the faint of heart, but I am hoping to to hear some success stories (or merely implementation stories) that will help the rest of us learn from the early adopters' mistakes. 
@mbushong
50%
50%
@mbushong,
User Rank: Moderator
12/10/2013 | 12:43:07 PM
SDN == OpenFlow?
2013 has certainly been a big year for SDN. The first 4-5 months were all about arguing definitions. Since then, there seems to be a marked shift in dialogue away from definitions to solutions. As part of this shift, the focus seems to be less on the specific protocols (OpenFlow) and more on the solution. I would be surprised to see the dialogue shift back to OpenFlow. That's not to say OpenFlow isn't important, but I don't think the supporting protocols are the primary point of concern for most people.

Rather than wide adoption, I suspect we will see a bit of disillusionment with SDN. The more likely deployment scenarios are single-vendor and in very controlled settings. Having SDN actually deployed will change the focus from concept to operations, and I think we will find out that things like monitoring and troubleshooting are not fully solved.

And if the real value of SDN is in workflow automation, people will find this year that cobbling everything together is harder than people think. The skill set gap in most shops will make early adoption hard. It's absolutely worth doing, but the first real deployments will be difficult for many. This is why professional services and systems integrators are salivating.

-Mike Bushong (@mbushong)

Plexxi
Cartoon
Slideshows
Audio Interviews
Archived Audio Interviews
Jeremy Schulman, founder of Schprockits, a network automation startup operating in stealth mode, joins us to explore whether networking professionals all need to learn programming in order to remain employed.
White Papers
Register for Network Computing Newsletters
Current Issue
2014 Private Cloud Survey
2014 Private Cloud Survey
Respondents are on a roll: 53% brought their private clouds from concept to production in less than one year, and 60% ­extend their clouds across multiple datacenters. But expertise is scarce, with 51% saying acquiring skilled employees is a roadblock.
Video
Twitter Feed