The flap over usage pricing, renewed last week by announcements by AT&T and TW, has raised again the question of how network infrastructure might respond to a new broadband world, one where unlimited usage no longer stimulates new apps. In such a future, operators would be accepting a role of bit-pusher, and the growth of broadband would no longer be unbridled. Many, myself included, think that the social-network and video bubbles would burst. Operators are somewhat cautious about the topic, but there are a few comments/directions we can relate and this is clearly a good time to be thinking about them.
So what’s a worthy approach? Operators’ number one strategy for the moment is to reconsider their content strategies. Video traffic is by far the greatest source of profit problems, and inside some of the current moves you can see the video issue weaving its threads. For example, we’re told that operators have concluded that there are no “hogs” or heavy users who are not heavy consumers of video. By setting usage cap points at the top end of what non-video users are likely to use, operators believe that they can focus their price-pressure remedy on the specific culprit of streaming.
CDNs are also a part of the video optimization picture, of course. In the past, CDNs have been largely promoted by content owners as a means of dealing with peering-limited delivery performance. But by staging content artfully closer to the access edge, you can unload the deeper metro infrastructure where operators say content is creating the largest pressure in wireline. In wireless, CDNs are helpful but can be made more compelling if you offload video from the 3G/4G cells, which could be done through expanded use of WiFi. That’s because cell congestion in wireless can happen before metro aggregation congestion.
Another strategy gaining credibility is the “usage-free” notion, something that is a perilous course from a regulatory perspective. The idea is to exempt applications or content sources from usage pricing, either because they’re your own or because the app/content source has elected to pay on your behalf. This seems to be at least a potential violation of the US neutrality principles, but because CDNs are explicitly exempt from neutrality it may be possible to concoct an architecture that would pass muster, and of course the whole order is in the courts on appeal anyway.
The next approach operators are looking at is pushing traffic down the stack. A three-layer architecture with IP at the top is the most expensive to deploy and support. Reducing the number of active network elements has a major value in cutting costs. Operators have been pushing for more fiber and less IP, and while vendors have blown kisses in this direction it’s very possible that things like OpenFlow might be used to augment basic optical switching/steering and create a network and even metro core with more optics than electronics, so to speak. This is what operators hope for in the long pull, but they believe they may have to wait until either optical players or other cost disruptors (like Huawei) figure out exactly how to do this; the big equipment vendors are in their view dragging their feet on optical consolidation.
If the current trends in revenue per bit are sustained, our model says that global network infrastructure spending would have to fall to less-than-replacement in the next couple of years, meaning that the total new assets being placed in service would be less costly than those being retired. In some geographies this is already true in wireline, but the thing that’s interesting is that we’re also approaching a point where retirement of older TDM stuff will largely cease to be a factor. Up to now, we’ve had a net gain for IP and Ethernet and optical even in the face of declining capex, but that’s coming to an end.