How and When SDN Rolls Out

[This post was originally written for and published on The Rayno Report.]

Software Defined Networking (SDN) has quickly spawned what appears in some respects to be a cottage industry of would-be disruptors to the more traditional networking approaches. With hundreds of millions of dollars in venture capital flowing into the space and dozens of infrastructure companies now vying to be the preeminent SDN vendor, how can anyone — customers or investors — predict who the breakout players will be?

 SDN is a How, Not a What

The most basic thing that potential customers and investors need to understand is that SDN reflects how solutions work, not what their fundamental purpose is. Whether or not something is SDN is almost immaterial to the role that the device or solution plays. Indeed, most common networking problems can be solved with both legacy solutions and their SDN counterparts.

Think about SDN like the type of engine in a car. For most buyers, it’s a detail that gets weighed along with how much cargo space there is and how many cup holders are available. For others, the distinction between a V6 and a V8 is hugely important. But even in the latter case, the type of engine typically serves as a proxy for how much power or acceleration the car has. The real objective is not the engine but what the engine provides.

Generalized Networking

Because the industry is focused on how SDN applies to generalized networking, there has been an over-rotation towards a vague set of use cases. These use cases are broadly applicable, which has the benefit of positioning SDN in lots of places. However, because they lack specificity, they don’t constitute a plan that people can readily say “yes” to.

The result is that there are an awful lot of companies pitching generalized SDN solutions that fit everywhere and nowhere all at once. The fear this evokes on the buy side is one of forklift upgrades or rip-and-replace solutions. The buy-in required to authorize a total change in direction is huge. It goes well beyond just the technical buyer. To displace an entire vendor, you frequently have to get political air cover, sometimes from someone as high up as the CEO.

Small companies won’t generally have the clout to unseat an incumbent in this type of setting, barring a solution that is an order of magnitude faster, cheaper and more scalable.

Specialized Use Cases

If generalized networking isn’t going to be successful, then what will be? SDN deployments (especially those early on) will be focused on very narrow use cases. It might be a specific application (HP and Brocade have focused on Lync, for example) or may be a very specific deployment scenario (such as lighting dark fiber between buildings on a campus). The more specific the use case, the easier it is for potential buyers to say “yes.”

The challenge, of course, is that no one wants to narrow their target addressable market (TAM). When you are a startup seeking funding, claiming your target addressable market consists of all Ethernet switching makes it easy to put up a multi-billion-dollar opportunity. If you narrow that to only those deployments where dark fiber needs to be lit up between data centers, you shrink your TAM.

What is Success?

However, success in the business world is not measured by TAM. Zero percent of a massive TAM is still zero dollars. Startups need to get cash into the company early. Those early deployments are important because they force an iteration of the product and provide success stories and customer references from which to build. Without these wins, it is easier to make continual, even meaningful, progress without ever making any money.

So Who Will Break Out?

The companies that will break away from the SDN peloton are the ones who focus on getting deployments in the real world. A narrow set of use cases trumps a massive TAM in this case even if the former is a lot less sexy.

Customers and investors should ask questions about the overarching value proposition. If the answers are always framed in the general case, it is likely that the solution is aimed at the general networking space. And aiming at something so large is not that different than aiming at nothing at all.

Instead, vendors should ask questions about target customers. Is there a repeatable use case that has traction among multiple customers? Or is every deployment a snowflake, unique to the set of conditions that created it? Breakout success requires scaling the business, and scaling means making things repeatable.

Finally, the standard questions about customer references should be present. This is not because of the specific references so much as what they indicate. Vendors that can push past the analysis paralysis that so many SDN buyers are facing typically have a solution that is scoped narrowly enough to get a “yes.” That, more than anything perhaps, is an early indicator of success.