Microsoft’s Xbox will now feature the ability to stream 26 FiOS channels and access (presumably the same network) content on VoD. The step isn’t a technological revolution because AT&T’s U-verse has been a streaming IP video service from the first (and it’s also available on Xbox, but in the same form; FiOS for Xbox is channel-limited for now). It may be at least a step in a kind of service revolution, though. The operators themselves are mixed on this topic.
Streaming content is often seen as a problem for “the Internet” but most streaming content is never really on the Internet at all. Popular content, even many YouTube videos, is cached near the access edge and often rides only a short distance to the user. Globally, operators tell us that “commercial-grade” content travels on the average only 11 miles from point of storage to point of delivery and that all of that distance is beyond the Internet peering point. CDNs, in short, control content traffic and not the Internet. The explosion in video content doesn’t really drive up “Internet” traffic, it drives up access network traffic.
So where’s the beef? Operators who don’t have a viable financial scenario under which to deploy FTTH are concerned that access congestion could truly kill their infrastructure. If you push glass to the edge you have a relatively unfettered upgrade path because fiber capacity is formidable. If you have copper, the problem is that you run up against hard technical limits. A third of operators say they can’t support IPTV on at least 50% of their plant.
There’s also a question of neutrality policy that concerns operators when they look at streaming video. By buffering you can make any IP path deliver content, but the buffering delay quickly turns off consumers. Getting someone to wait even a full minute for video to start is a challenge, and even ten minutes of buffering (which no one would wait for) on a 60 minute show can produce stutters in the stream if there are significant variations in the traffic congestion along the path. Thus, video should in theory have some sort of priority. While neutrality rules don’t necessarily rule out video priority, they do leave questions on how it could be paid for and whether priority paths for video could be “gamed” for use in other applications.
Interestingly, operators weren’t that worried about the classic issues of video revenue per bit on streaming services; they see more of a problem with “non-commercial” video like YouTube that can pull video further through their infrastructure and thus create more capacity problems in more places. The alternative for them is to cache more video, and for customer-loaded material it’s hard to do that except reactively. Mobile video creates issues for them because often it’s stored and streamed in a form that can’t be displayed on the device anyway (it’s down-sampled, in effect) but that still uses up the full repertoire of bandwidth along the path. That means transcoding, and when access or downstream points get congested it means adaptive transcoding. You’re getting the picture here.
So why is Verizon doing this, and why has AT&T already done it? Answer; because they can. AT&T’s situation is easy; U-verse is already IPTV. Verizon’s situation is relatively easy too; FiOS has the capacity. More to the point, the incremental cost of supporting streaming to Xbox or something else isn’t high enough to invalidate an access business model that meets ROI goals. Add channelized TV to broadband in a sensible way and you can still make money where customer densities are reasonably high. The moral is that every market is different, and announcements of services in Japan or Germany or even the US Northeast don’t mean that everyone will eventually see these services.
On that matter, we’re looking at the enormous FCC “Connect America” order now, the replacement for the old Universal Service rules with an excursion into intercarrier compensation. We’ll be blogging on this for our TMT Advisor premium blog when we’re done.