How Mobile-Model Changes Could Impact SDN and NFV

Mary Meeker is probably best known for her role as advocate-in-chief of the dotcom bubble, but she’s produced some respectable research since then, and her latest has some interesting points, particularly when considered in light of SDN and NFV evolution.  In fact, she suggests that there may be an emerging set of missions for both that are rarely considered when talking about how networks will change under business pressure.

Meeker’s key point is that penetration of mobile devices is already almost tied with that of TVs globally.  While at present only a third of mobile devices can be called “smart”, we’re heading for a world where smart mobile appliances are the most common of all communications devices, and thus where service trends and infrastructure requirements for mobile will dominate.

Most of you already know that mobile dominates infrastructure spending, but this dominance is based not on traffic or population of users but on profits.  Over the last couple of years, mobile profits have been falling and returns on mobile infrastructure have been falling too, to the point where they are close to being aligned with wireline.  One question this trend created was whether there would be a re-balancing of wireline and wireless investment created by this leveling of return, but it now appears that the commoditization we see in mobile is also creating a much larger mobile market.  Mobile is winning.

When you look at this point, you see an interesting question, which is that of mobile usage.  Not usage in terms of quantity of traffic but type of traffic; what are people doing with mobile devices?  Video is always ballyhooed, but the most common mobile apps are Facebook, Twitter, and LinkedIn.  One question this raises is whether we might see mobile services aimed at the population of mobile users who don’t care for stupid pet tricks, and how they might be served best.

Carriers are already looking at ways of creating mobile services that focus on non-video traffic, and I’ve heard that there are some quiet discussions in DC on how regulators could support this notion without creating a neutrality flap.  One way would be to offer neutral service but lower prices for those with very low monthly usage, which would favor those not interested in streaming, and I’ve also heard this is under consideration already, particularly for prepay carriers or MVNOs.

But perhaps the most interesting question is the possible tie between the mobile trends and Google’s publicized exploration of low-orbit satellites as a means of providing global Internet services to “under-served” or “undeveloped” populations.  I wonder whether Google really sees a huge ad market in this kind of venture, or whether they might instead see a chance to offer a satellite-based service that would support low-volume social networks?  Would Google, if they offered a service like this, be considered a “carrier” at all, and would they have to share anything with competitors?  Could a Google+ service, Gmail, and other G-stuff be linked to a simple handset and offered anywhere as a kind of “social smartphone?”

How useful this would be depends on the behavior of the 3.6 billion mobile phone users who don’t have smartphones.  Obviously they’re not streaming video today.  Would they elect to step up to video capability at a higher price, or convert to a “socialphone?”  If the latter is more likely then we’d probably have more growth in customers than in traffic, less focus on SDN in Evolved Packet Core applications and more on pure IMS, and even less focus on NFV for mobile.

What?  We’ve heard all the stuff about mobility and NFV, so it may seem strange that mobile might not be an ideal NFV app, but the truth is that NFV and mobile aren’t necessarily perfect partners in a benefit sense.  The majority of IMS work being done today is focusing not on NFV IMS but on cloud IMS, for the logical reason that IMS is a multi-tenant application by nature and so doesn’t require multi-tenant specialization in deployment and orchestration of the kind NFV provides.  Automating orchestration and operation is most valuable for services that are not naturally supported by multi-tenant software.  IMS is loaded and run for all customers, not loaded for each customer or call, so there’s not much dynamism there.

The effect of the cloud-versus-NFV distinction could be significant in that it could bring about a power shift in “NFV” away from the ETSI NFV ISG and to other bodies.  Orchestration of software components is less an issue in monolithic multi-tenant apps but management orchestration may be more important because what dynamism there is comes from responding to changes in load or operating state—management functions.  But it’s not completely clear just where the impetus for management orchestration will come from as yet.  For IMS, bodies like the 3GPP could surely do something, and OASIS has a cloud orchestration model (TOSCA) that would serve both deployment and management orchestration in theory.  The TMF, at this week’s Nice meeting, is presenting a number of Catalysts (demonstrations or PoCs) that could shed light on management orchestration practices, but so far I don’t have public material on the submissions for me to comment on.

Any fragmentation of the “NFV” space and disconnection of SDN from mobile services could create both issues and opportunities.  Mobile is the largest single infrastructure investment focus as I’ve noted, and that means that if it’s on the table there’s going to be pressure for a systemic solution to both SDN and NFV.  Absent mobile we could see smaller applications justifying point solutions, particularly if some common management integration model can be found to harmonize OSS/BSS linkages.

Of course, none of this matters if there’s no real shift in mobile strategy.  If mobile EPC evolution is strong then it drives SDN systemically.  If services that exploit context grow in importance, then the inherent dynamism of these services could still create a model where orchestration of both deployment and management will be essential.  The question is whether contextual services are part of mobile infrastructure or part of the OTT world.  Apple or Google may be the key to this, or Facebook or Amazon.  Any of these companies could either become MVNOs or decide to deploy their own networks, and if they do they may change the mobile game as they go after those over-three-billion mobile users who’ve not committed to smartphones…yet.