• 08/07/2014
    6:59 AM
  • Rating: 
    0 votes
    Vote up!
    Vote down!

SDN: You Can't Argue Against It

As executive director of the ONF, Dan Pitt hears all the excuses about why organizations are waiting to adopt SDN. Here he explains why they're wrong.

As an organization dedicated to software-defined networking (SDN), we usually focus on the positive -- how many end-users are adopting, planning for, or considering SDN. These increasing numbers are proof that the efforts of the Open Networking Foundation (ONF) are making an impact. Yet, according to the "Seventh Annual State of the Network Global Study" by Network Instruments, 68% of respondents still have no plans to deploy SDN. If your company is among them, you are three years late to the game.

I am very familiar with the rebuttals to that statement; I hear them often as the executive director of the ONF. If one of the below responses is the reason your company has held off getting on board with SDN, here's why you should reconsider: 

"We're not three-years late. SDN is up-and-coming technology."
SDN has been referred to as "up-and-coming technology" since well before ONF was launched in 2011. It is still considered a new innovation, but companies need to be participating in the conversations that have been taking place in order to stay up to date. There are hundreds of actual end-users leveraging SDN within their networks (Google, NTT, Genesis Hosting, Stanford University, Pacnet, and Kanazawa University Hospital, just to name a few). SDN is being realized, and it is only a matter of time before it becomes mainstream. Companies that do not begin engaging in and planning for the transition now are going to be at a disadvantage among the competition, at risk of being left behind.

"The SDN products that we need aren't in the market yet."
I think companies would be surprised to learn how many SDN product lines are now commercially available from vendors of all sizes, from leading companies to promising startups, focusing on solutions for cloud, telecommunications, data center, wireless, campus, and more. We have a substantial product directory on our website. These fall into three main categories: software-based OpenFlow switches (like those based on OVS), hardware-based OpenFlow 1.0 switches, and software products in and above the control plane.

I am aware that the market is waiting for more hardware-based OpenFlow 1.3 switches, which will emerge this fall in much larger numbers. But there are so many products available for at least lab trials or small deployments that you have no excuse for not ramping up your learning based on these.

"SDN requires new skills that our team does not currently have."
SDN does require some new skills, and new processes, but the best time to begin acquiring those skills is now. An increasing number of venues offer training. Education and skills development will help you preserve your biggest investment -- your human resources. If you invest in them, they will love you back.

"Our networks cost us a lot of money, and transitioning to SDN would be too expensive."
We understand that networks are serious investments for companies, and we want network operators to preserve their investments, so it's pretty amazing that so disruptive a development as SDN can be introduced gradually, without a forklift upgrade. However slowly you start, the benefits can accrue immediately, reducing capex and opex and creating new opportunities for service differentiation and revenue growth.

"There's no straightforward, tested way to migrate to SDN."
It's good that there is no single way to migrate to SDN, thanks to the infinite number of ways operators can exploit SDN. But newcomers can benefit from the experiences of those who have at least begun the transition. I recommend engaging in conversations with companies that are implementing or have deployed SDN to learn what works, what doesn't, and how the process can be improved. ONF's Migration Working Group explores and documents how companies have made the transition, and develops migration tools and metrics. 

"It is not a priority for us right now."
It needs to be. Given the importance of the network to the IT function and a company's operation, if the network is out of date, so is the company. Not too long ago I was challenged by a major network operator who asked how I was going to compel his company to adopt SDN. I responded that I was not going to compel him to do anything -- his competitors would, by offering services he could not offer at prices he could not match. If you aren't at least experimenting with SDN already, it's time to start. 

In what ways has your company embraced SDN? I'd love to hear about your motivations, as well as the challenges that you overcame, in the comments below.


Resistance to change

Dan, all your arguments in favor of SDNs are strong and valid. Ultimately it is the human resistance to change working at corporate level.

While changing to SDN could be benefitial for most organizations both CIOs and CFOs are usually reluctance to invest in new technologies until it is absolutely necessary.

Re: Resistance to change

For the veterans out there time will tell with SDN and NFV. Remember all the "ATM will change the networking world" chatter some 15 years ago. Some jumped in and ended up regretting it. Others waited for the technology to mature and settle down to see if it was worth pursuing and were glad they dodged a bullet.


Remember snmp and asn was supposed to do all this for us. Provision, monitor and control our networks though mibs. Well how did that ultimately workout?

The trick is to determine what not to learn/embrace and only experience helps with that thinking. I follows SDN/NFV but just on the sidelines right now. Let the dust settle on the technology and vendor fronts before committing additional time.

Re: Resistance to change

Jeffrey, I think your comment is right on target. I do believe SDN has substantial benefits, but we need more standarization and competition.

When SDN becomes the norm, the reluctance will disipate.

Re: Resistance to change

There will always be people that are reluctant to change. As soon as they notice that the benefits outweigh the initial annoyances, everyone will end up coming around. Besides the IT industry should easily get used to quickly changing directions like these.

SDN migrations

I think perhaps once there are more enterprises talking publicly about how they adopted SDN -- and the results -- it will encourage others. It seems that most of the SDN case studies we hear about right now involve service providers, large Internet companies or educational institutions.

Re: SDN migrations
Marcia, I think that's a huge part of it -- no one wants to be the early adopter making all the mistakes that other companies will later learn from. Dan says in his blog that SDN is being used today, but when he goes on to list Google, NTT and Stanford as examples that only goes to show how early in the adoption phase we are. Tell me about some low-profile companies with average employees, please.