T-Mobile has been shaking up the mobile industry with its no-contract-and-phone-upgrades plans, and they’ve worked to gain post-pay customers for the carrier, not a first but certainly a recent first. The problem is that this “good news” really isn’t more than a “transient good” because this model isn’t sustainable.
Suppose you’re in the milk business. Milk is tough to differentiate on features and you don’t want a milk price war, so you decide to offer people a free glass with their milk, in return for their committing to buy from you for the next six months. Your competitor will quickly match that, and maybe raise it to a two-glass set. Next thing you know you’re offering a complete set of dishes and a table, and competitors are quickly matching every stride you take. All that happens is that everyone’s profit margins go down the tube.
The answer, as we can likely see from this example, is to find a way to feature-differentiate what you’re actually selling. As simple as this may seem, the telecom industry has done a truly awful job of facing the obvious, for a variety of reasons. Biggest of these is that it takes time to find service differentiation, and operators are like every other business these days—focused on delivering the next quarter. And network equipment vendors are in the same boat.
I talked to some sales people at some of the giant equipment vendors over the last month, and one thing I found interesting was that all of them said that their company was committed to strategic selling. The problem was that they were also dedicated to meeting quota for the current period, and you get paid for the latter and not for the former. Hint to sales managers: Sales people sell what you commission them to sell, not what you tell them to sell.
Logically, the solution to this situation would be for vendors to build future service growth into current product sales, so that new stuff was suitable to build differentiation and ARPU. The problem in that picture is that it’s extremely difficult to do much in the way of future-proofing at zero marginal cost. With Huawei out there cutting margins for everyone as it is, who would want to go into a head-to-head with Huawei with a five or ten percent cost disadvantage created by adding features need in the future but not in the present? Sounds like a gold-edged milk cap to me!
This is background for the whole SDN and NFV thing, IMHO. Underneath it all, what operators are trying to do is transform their business model in a way that doesn’t require them to pay too much forward—they have to control “first cost” and return on infrastructure. In the surveys I do, I see a clear signal that they want to transform in two phases—a phase that delivers next-gen infrastructure justified by current-period cost reduction, and another phase that exploits that infrastructure to create differentiated services.
SDN was a failure as a strategy here. First, it doesn’t deliver revolutionary savings. Second, it is a connection-plane model so it’s not going to create higher-level feature value-add, and at the end of the day pushing a bit with Technology A isn’t much different to the bit-pushee than it would be with Technology B. Now the torch has been passed to NFV, and NFV has the advantage of being primarily an IT technology rather than a network technology. You can say that it’s a “virtualization” approach, but it’s really more a software approach. With NFV you say goodbye to boxes and run applications that replicate the features those boxes provided. If you demand sanctioned, standardized, interfaces and functionality for each virtual element then it’s easy to play best-of-breed and lowest-price with competing vendors, so your costs are managed. That’s where the first-phase benefit comes.
For the second phase, you’ve got to be more of a visionary. The cloud is the future of IT, so the cloud is the future of NFV. If NFV can create an architecture that can deploy and manage services created by cooperative virtual functions, it can deploy multi-component cloud computing applications, and hybrids of these two things. This means that the operator gets a future-proof service layer as a byproduct of getting the benefits of capital cost reduction. No first-cost risk any more.
Vendors, of course, are largely un-thrilled with this model. While it’s possible that it might create as much or more trouble for Huawei as it does for the rest of the field, it creates a whole new host of competitors and it also eliminates the proprietary lock that appliance-based services tends to create. At any rate, the vendors are all very careful to guard the boundaries of whatever NFV they’re thinking about in an attempt to protect what they have while chasing what their competitors might have. But with NFV established as a common goal of network operators, can vendors ignore it? The first guy who conforms gets an advantage, so in the end the vendors will follow along just like the other operators followed T-Mobile’s lead in plans and phones.
The challenge for NFV in meeting its goals, and those of the operators, is partly technical and partly procedural. On the technical side, the management of this new “supercloud” is something out of the twilight zone in terms of issues and changes. Attempts to solve the management problem of NFV infrastructure through traditional means risks compromising opex and the whole NFV business case. On the procedural side, the big question is whether the ISG can sustain an open model at the three layers of NFV—the virtual network functions that replace appliances, the ecosystem that deploys and manages them, and the infrastructure they run on. Implementations will be the final proof of this point, and we’ll likely see many more coming this fall.