Several recent online articles have pointed to a lack of enthusiasm on the part of network operators for deployment of SDN (see here and here, for example), leading to discussions online and at the upcoming OFC Executive Forum  on whether or not SDN can truly be a business for carriers and vendors outside of the few Google-sized datacenter operators in the world. The answer is still likely yes, but on a level that makes much more sense than the original overblown promises.

Initially, SDN was proposed as a way to convert entire networks to low-cost, multi-purpose white boxes that could be programmed to make the entire network flexible. Despite statements from AT&T, that promise will probably not be realized on a large scale in most networks. There are large scale SDN deployment examples such as the PacNet (now Telstra) SDN-enabled network, but most are largely single-vendor solutions that do not extend beyond bandwidth-on-demand types of services. While bandwidth-on-demand is a big part of SDN, SDN is not only bandwidth on demand. The promise of SDN is much larger, a fact that has been touted by vendors for years – which is perhaps the reason for the recent pushback from network operators.

When a vendor approaches a carrier with stories of network transformation and lowered operational expenses, the first question is inevitably how much that transformation will cost in dollars, time, re-training personnel, and re-designing operations. Most carriers are legitimately overwhelmed by a network transformation story, especially since most such stories are currently only half baked with significant pieces still in development. Unless the carrier builds the entire network from a single vendor, solutions are still in the emerging stage, and any solution that claims to be SDN while requiring that all of the parts come from a single vendor is just an EMS. Some controller vendors would argue that complete solutions can be built today, but the reality is that no controller can operate all of the currently “SDN-enabled” equipment on the market – much less the legacy equipment already installed – without significant custom software development. This “we will transform your network” approach gets meetings with carriers, but runs into walls of reality very quickly.

An alternative approach, and one that seems to be gaining traction with smaller carriers and a few larger carriers, is to approach SDN from the edge of the network with targeted applications. Whereas a network redesign may be intimidating, providing a more agile access network, new Enterprise services, or SD-WAN capabilities can be easier to absorb, test, and rollout on a limited basis. That is why announcements like the recent Verizon/Viptela news release are more interesting that announcements of theoretical network transformation from vendors .

Read more closely the article about Orange downplaying SDN, and you’ll notice that they did not say that they will not deploy SDN, only that the business case must be made that deployment will make money. Smart vendors with longer experience have been making this case from the beginning – SDN is not just a way to “transform your network”, it is a way to offer new services that will make money. And most of those services will start at the edge where equipment changes more rapidly than in the core.

In a previous life on the marketing team at a telecomm vendor, we used to say, “we are building the SDN bricks that you need to design your SDN house.” The point was that network operators should understand that the promise of SDN is coming, but that the infrastructure needed to realize all of the benefits will take a while to be deployed. In the meantime, it is in the best interests of the network operators to install equipment and software that is ready when the rest of the infrastructure is available. That strategy – build SDN-ready equipment and software designed for the future – is the most reliable way to make money on SDN, at least for the near future.

There is one other way to make money on SDN as a vendor. Build up your software team, develop impressive SDN-themed software, and get yourself purchased a la Cyan and Overture. But one has to believe that those deals will be much harder to secure as the SDN market matures.