Why Many Networking Professionals Will Resist Software-Defined Networking

In the long run, I think software defined networking (SDN) is destined for tremendous success, not only at massive cloud service providers, where it already is finding favor and increased adoption, but also at smaller service providers and even — with time and perseverance — at enterprises.

It just might not happen as quickly as some expect.

Shape of Networking to Come

In a presentation last autumn at the Open Networking Summit, Nicira co-founder Nick McKeown asserted that SDN would shape the future of networking in several key respects. He said it would do so by empowering network owners and operators, by speeding the pace of innovation, by diversifying the supply chain, and by delivering a robust foundation for programmability predicated on a standardized forwarding abstraction and provable network properties.

On the whole, McKeown probably will be right, and his technological reasoning seems entirely reasonable. As in any market, however, the commercial appeal of SDN will be determined by human factors as well as by technological considerations.

The enterprise market will be the toughest nut to crack, though, and not only because the early agenda of SDN, as defined by the board members of the Open Networking Foundation (ONF) and others, has been focused resolutely on providing solutions for the largest of cloud service providers.

Winning Hearts and Minds

Capturing enterprise hearts and minds will be difficult for SDN, and it will be hard not just because of technological challenges, such as backward compatibility with (and investments in) existing network infrastructure, but also because of the cultural milieu and entrenched mindset of enterprise networking professionals.

I’ve written before, on two occasions actually, about how human and institutional resistance to change can strongly inhibit the commercial adoption of technologies with otherwise compelling credentials and qualifications. Generally, people fear change, especially when they suspect that the change in question will affect them adversely.

And make no mistake, software-defined networking will inspire fear and resistance in some quarters, enterprise networking professionals prominent among them.

Networking’s Cultural Artifacts

Jennifer Rexford, professor of computer science at Princeton University and a former AT&T Research staffer, wrote that one of her colleagues once observed that computer-networking people “really loved their artifacts.” Those artifacts probably would include the many distributed routing protocols that have proliferated over the years.

Software-defined networking wants to loosen emotional attachment to those artifacts, just as it wants to jettison the burgeoning bag of protocols that distinguishes networking from computer programming and other disciplines.  But many networking professionals, including those in enterprise IT departments, see their mastery of complex protocols as hallmarks of who they are and what they do.

Getting the Network “Out of the Way”

Yet there’s more to it than that. Consider the workplace implications of software-defined networks. The whole idea of SDN is to make networks programmable, to put applications and those who program and manage them in the driver’s seat, and to get the network “out of the way” of the sweeping virtualized progress that has enveloped all other data-center infrastructure.

To survive and thrive in this brave new virtual world, networking professionals might have to become more like programmers. From an organizational standpoint, even though there are compelling business and technological reasons to adopt SDN, resistance from the fraternity of networking professionals will be stiff and difficult to overcome.

In the realm of the super-sized data centers at Google and elsewhere, this isn’t a serious problem. The concepts associated with “DevOps” and with thinking outside boxes, departmental and otherwise, thrive in those precincts. Google long has eschewed the purchase of servers and networking gear from vendors, and it does things its own way. To greater or lesser degrees, other large cloud-service providers now dance to a similar beat. But the enterprise? Well, that’s a different animal altogether.

Vendors in No Hurry

Some of the new SDN startups already are meeting with pockets of resistance. They’re seeing cleavage — schism might be too strong a word, though maybe not — between cloud architects and server-virtualization specialists on one side of the house and network professionals on the opposing side. The two camps see things differently,with perspectives and priorities that are difficult to reconcile. (There are exceptions to the rule, of course, with some networking professionals eager to embrace SDN, but they currently are in the minority.)

As we’ve seen, the board of directors at the Open Networking Foundation (ONF) isn’t concerned about how quickly the enterprise gets with the SDN program. I also would suggest that most networking vendors, which are excluded from the ONF’s board, aren’t in a hurry to push an SDN agenda that features logically centralized, server-based controllers. You’ll see SDN from these vendors, yes, but the control plane will be distributed until such time as enterprises and service providers (not on the ONF board) demand otherwise. That will be a while, I suspect.

Deferred Gratification

We tend to underestimate resistance to change in this industry.  Gartner devised the “trough of disillusionment”  and the technology hype cycle for good reason. Some technologies remain in that basin longer than others. Some never emerge from what becomes a bottomless pit rather than a trough.

That won’t happen to SDN.  As I wrote earlier, I think it has a bright future. Don’t be surprised, though, if the hype gets ahead of the reality. When it comes to technologies and markets, our inherent optimism occasionally is thwarted by our intrinsic resistance to change.

8 responses to “Why Many Networking Professionals Will Resist Software-Defined Networking

  1. Interesting article. Couple quick points:

    – Her name is Jennifer Rexford, not Wexford
    – It’s kind of ironic that this is the case, given some of the early work on SDN (e.g., Ethane) was specifically focused on enterprise networks.

    • Thanks for catching my typographical error on the misspelling of “Rexford.” I’ll blame it on allergies, which were fogging my brain when I wrote the draft. 🙂 I’ve corrected the error.

  2. Risking to appear naive, I would say that the more obvious is the value a particular technology brings to the table (especially to the table where those who are holding purse strings sit), the quicker it gets adopted.

    If SDN can be convincingly shown to affect the bottom line for the “mass business”, I have little doubt that its adoption (or at least encouragement to investigate) will spread much quicker that it would otherwise. Just look at VoIP.

    However, what I’m not convinced about is whether SDNs in their present shape have that “golden touch”, because they’ve been conceived to solve problems that are not on the top priority list of the mass enterprise. I wrote about a similar issue recently: http://telecomoccasionally.wordpress.com/2012/02/20/mid-market-innovators-dilemma/ , and it is also applicable to Enterprises directly (while my blog post centered around Service Providers).

    I am not sure how to solve this, but I think the situation is literally charged with opportunity, and it just needs a right combination of insight and capability to crack it.

    • Thanks for the comment. I don’t disagree with your first point, and I agree with your second.

      On the first point, though, I think that networking professionals can definitely run interference on any technology they don’t want to adopt. I’ve seen it happen. Eventually, if the ROI of SDN becomes obvious to those who hold the enterprise purse strings, those objections can and will be dismissed or overruled. However, that won’t happen until the business cases have been validated clearly (and perhaps repeatedly) in relevant reference accounts. Remember, too, that legacy networking vendors aren’t going to go down without a fight, and they’ll often make common cause with the networking contingent inside the enterprise.

      It’s certainly possible that enterprises and their legacy vendors will opt for a form of SDN and network virtualization that favors a distributed control plane rather than the logically centralized control planes proffered by some SDN startups.

      • > network virtualization that favors a distributed control plane

        As I commented elsewhere, I feel that there’s merit in a hybrid approach to SDNs, where network edge functions (e.g., ingress classification and forwarding path association) are performed using SDN mechanisms, while core forwarding is based on “traditional” distributed control plane.

        IMO, such approach might be acceptable to legacy vendors and would give them a ticket to the SDN party while not jeopardising their established position too much.

  3. Well, maybe. But I was struck by an interesting observation in another blog earlier today:

    “Usually early adopters are a tiny portion of the market. What’s so interesting about the cloud, about the Googles, Facebooks and mega data centers is that they’re early adopters and a major portion of the market.”

    From http://gigaom.com/cleantech/low-power-is-the-new-black-for-data-centers/

  4. Pingback: SDN in the Enterprise: Like Oil on Water | Silicon Loons

  5. Stuart Miniman

    Excellent discussion Brad – while the typical enterprise IT organization would love to drive down costs and improve agility, it is FEAR of outages or failed projects that drives most administrators. Saving a little money won’t get someone promoted, but if the network goes down, they could be out of a job. The Googles and Facebooks of the world have to adopt the new technologies to enable their businesses, enterprise environments might have difficulty with the current deployments, but they don’t yet have a gun to their heads to jump on SDN. Great trend, but like most technology adoptions, we’re likely to say that we’re disappointed at what adoption takes place in 1-2 years and amazed at what happens in a decade.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s