The holiday season is always dominated by consumerism, but it should be pretty clear to everyone that networking itself is increasingly dominated by the consumer. I think that we’re headed very quickly for a time when the consumer essentially funds all public networking, creates the design paradigms and the economic trade-offs. Along the way, though, we’re facing some potentially significant hurdles and shifts in the course.
The Internet has already made public IP infrastructure the basis for public networks, though of course that infrastructure tends to be less homogeneous than many see. Ethernet is a smarter edge strategy, for example, because most consumer services will haul traffic to either a metro off-ramp or a metro cache/server farm. You don’t need a lot of connectivity to get to one place. Still, the Internet has won IP a victory at the service layer, where the IP address space is the only framework we could expect to see in the network of the future.
This month, we’re heading to a kind of financial watershed with public network services. There’s been a surge of growth in online services funded by advertising, but advertising represents only a fraction of the money needed to fund a public network, and recent legal disputes (on Interclick’s history-tracking, for example) show that advertising-related sites are pushing the limits of public and judicial tolerance in a quest to tie up those limited dollars. Ultimately people have to pay for stuff to fund a three-trillion-dollar-worldwide industry like networking. The FCC is likely to set the boundaries of where pay works and where it doesn’t in its December 21st order on net neutrality. But whatever they do, there’s no turning away from the fact that advertising isn’t ever going to fund the public network, so something else has to.
Consumers would love a free Internet, just like they’d like free automobiles, homes, or cheese. That doesn’t make the concept practical, even in a political climate where give-aways are the rule and not the exception. We’ve taken free-ness about as far as we can at this point; even Google I think understands that it has to move from being totally ad-driven to having some set of for-pay products and services.
What the FCC’s order will do is establish the legal framework for an Internet that’s cooperative in a broader way than at the pure connectivity level. What’s needed is the same today as it was back in the mid-90s when I participated in an attempt to bring financial order to the Internet by creating a formalized mechanism for peering and settlement that included QoS. We have the technical means to do what’s necessary, but we don’t have regulatory air cover. The question now is whether we can get it.
Genachowski’s attitude on net neutrality appears to have undergone a transformation, and at the same time the Comcast/Level 3 settlement seems to open the door for settlement between content providers and CDNs and access providers. Any settlement at all here would be better than we have, but Comcast/L3 doesn’t go far enough. It comes down to a question of whether the relationship is “peering” or “transit”, and neither of these concepts goes far enough because both are simply different ways of viewing the permitted traffic balance. There’s still no QoS-based settlement, and without that the Internet can’t provide pan-provider quality of experience. There will be a transformation of investment and a transformation of Internet architecture if we can’t settle QoS-based relationships across ISP boundaries. Such limitations favor investments in caching over interconnection, and favor larger and larger players to create fewer and fewer inter-provider boundaries. We may start to hear some details on the forthcoming order leaked this week, in advance of the meeting. Pay attention; it could be critical.