The service-provider SDN market —hardware and software combined — was about $840 million in 2013 and should grow to $15.6 billion in 2018, according to a report released this week by ACG Research.

Hardware accounts for about 80 percent of those sums, which makes sense when you consider the expense of hardware compared with software. “Our thinking is that, from a value-added-stake point of view, the underlying network elements become even more valuable if they are enhanced to work with the control element,” says Paul Parker-Johnson, the analyst who wrote the report.

Note that the numbers don’t include the enterprise market. They apply to service providers, including large-data-center owners such as Google and the big data centers of carriers such as NTT or AT&T.

Here’s how ACG’s 2018 forecast breaks down. The numbers reflect only “live” SDN deployments — cases where service providers would be using the technology, not just installing it for future use.

 

 

“IP services” refers to the IP service edge of a service provider’s network, “usually where things like VPN technology marries with pure IP routing or Layer 2 transport or multicast Ethernet transport,” Parker-Johnson says. “In that area, things like path computation or service chains depend on the service provider’s actual offering. That one’s different enough, in our point of view, that it deserves its own little bin.”

Data centers for cloud IT and applications will actually take up more than 50 percent of the service-provider SDN market through 2016, Parker-Johnson believes. But after that, he expects SDN deployment to accelerate as the technology becomes more of a cookie-cutter implementation.

“We’re going to be in a period of testing and incubation — experimenting with the other domains. When we get out to two or three years from now, the trust and ability to prove out transport and IP services will get a lot of traction,” he says. As that happens, the spending in the non-data-center categories would grow quickly due to the vastness of carrier networks, he says.

You might recall SDNCentral reporting a $35 billion SDN market forecast for 2018. That analysis doesn’t compare directly with Parker-Johnson’s numbers; it wasn’t limited to service providers, and it included SDN-ready equipment, sales that represent customers preparing for SDN even if they wouldn’t be using it immediately.

Parker-Johnson expects that SDN enabled market, combining live and potential SDN deployments, to be around $25.5 billion for service providers in 2018.

Controllers Can Make Actual Money
The SDN controller is becoming an open-source animal, with free options joining the market every week, it seems. (Freescale apparently has one, for instance.)

But that doesn’t mean the market size for controllers is zero, because controllers with integrated applications should be a viable market, Parker-Johnson says. “I think one category is going to be controllers with specialty service applications” such as workload optimization of some kind, he says.

The market for controllers and applications combined could be on the order of $3 billion a year by 2018, he thinks. (And to reiterate: No, that’s not $3 billion worth of applications and $0 worth of controllers.) This is a market being created by the need to build an SDN tier and a separate control plane inside the network — a need that service providers didn’t previously have. “We’re monetizing a whole bunch of functionality that’s going to change the model,” Parker-Johnson says.