AT&T, SXSW, and How Virtualization Can Make a Great Time Even Better
March 24, 2015 Leave a comment
- For years, network equipment vendors have provided support services to help improve network capacity and user experience at large events.
- AT&T routinely rolls out specially designed mobile base stations to large events to help handle excess demand.
- NFV solutions related to vEPC and vIMS, and perhaps even virtualized BSS, can theoretically help complete the experience by supplementing the core as these excess demands become increasingly sophisticated.
When people ask me what I do at Current Analysis, I generally respond that I’m the “vendor services guy.” You know, network optimizations, consulting, managed network ops… those are the topics that make up my “beat.” As part of this coverage, I look at engagements that are commonly referred to as “special event services.” In a nutshell, this is where vendors such as Alcatel-Lucent, Ericsson, Huawei, and Nokia help their customers prepare for large festivals and other gatherings by adding extra base station capacity, as well as some network optimization to cope with the additional demand that a flood of users in a concentrated space can place on a mobile network.
Now, not to go all 1989 on you all, these services have typically been more concerned with call completion than truly holistic end-user experience. I can personally attest to this because I also live in Austin, and we get a lot of special events in our neck of the woods. Music festivals, Formula 1 car races, and X Games are all annual stops on the special event circuit. Invariably, the data service falls over as soon as the planes start landing.
So, when I got word that AT&T was taking matters into its own hands for the 2015 edition of South by Southwest (SXSW) by rolling in a fleet of what they call “COWs” (cell on wheels), and even a Super COW, specifically aimed at improving the festival goer’s mobile data experience, it was something I had to see for myself.
Yes, I agree, it kind does look like the Death Star. But, in retrospect, it appears that wheeling in the proverbial big guns was necessary. Per AT&T, its network alone carried 37 terabytes of traffic during the event, which was nearly double what it did last year. And, at least from my point of view, I have to give credit where it is due. My browsing experience, while in the midst of scores of thousands of multiple-device toting technophiles at SXSW Interactive, was excellent. By my unscientific method, I’d say it rated as about as good as an in-home WiFi experience… which is to say, one helluva lot better than most downtown streets during a typical non-SXSW rush hour.
However, it also got me thinking. While loading a map or looking to find out who was playing where and at what time was nice and quick, what about trying to do some heavy duty stuff that involves more complex actions throughout the network? What about broadcasting a Meerkat or ordering some international roaming data or buying some tickets online (I actually did the latter and the network experience was OK, but not great). What about the near future, when an event like SXSW would be a contextual marketeer’s dream? That, I suspect will require more than just a few strategically placed Super COWs. That will probably require some serious core network, and even back office, muscle.
Enter Network Functions Virtualization?
In a pre-Mobile World Congress briefing with analysts to preview his MWC keynote, AT&T’s John Donovan noted that the company wants to have up to 75% of its network functions virtualized by 2020. That’s a lot of functions. He also mentioned that it would include moving COs to a “data center model” (i.e., next-gen CO). So, by Mr. Donovan’s account, for a company like AT&T, being able to spin up extra mobile core capacity under this virtualized telco data center model should be routine.
But what if you’re not AT&T (or one of the select few other SDN/NFV first movers)? To a services guy like myself, it sounds like a potentially neat take on the old SES model. Hosted instances of virtual EPC and/or IMS? Maybe the ability to turn up some extra BSS capacity? As vendors are beginning to look at ways of hosting VNFs closer to the network edge, there’s no reason why some of this capacity couldn’t even be sited along with the Super COW… or even a regular COW.
While I don’t know exactly what that service engagement would look like right now, I do suspect it is where the model is going to have to go in order to provide a truly special “special event service” in the very near future.