Mediating IP Peering and Transit (Part I)

I think I've given you a good overview of what I mean by mediation, before. In essence, mediation not only means offering Quality of Service but also throttling applications which are ill-behaved or should not be consuming too much bandwidth during congested usage periods. Some in the industry call this network bandwidth and traffic management. It means that we don't want the teenager down the street running BitTorrent to be hogging all of the neighborhood's bandwidth while we're trying to download an email with a major customer presentation we have to prepare for the next day's business meeting.

These are a fair and reasonable business practices by broadband Internet Service Providers (ISPs). From InterStream's point of view, we simply wish ISPs would be more honest in relating to their customers what they're actually buying each month with their broadband access fees. This blog post explains why that is both important and possible.

Congested Broadband and Consumer Terms of Service

As Internet video congests the network, ISPs need to figure out which applications should get which service levels. The challenges and solutions are as inevitable as what we find on the average California freeway today. In my home town, Pleasanton California, it was just announced that their going to be putting up local express lane toll-roads to bypass the congested Interstate 580 corridor. In some locations of California freeways, semi-trucks are not allowed to use the road during rush hours. These are all natural solutions to the problem of congestion.

Some now estimate that the average broadband consumer will be downloading 50 times as much data per month as they do today. The growth of Internet video consumption is driving a broadband revolution. I certainly, don't mean this in a Malthusian sense. (Malthus was a pretty lousy economist!) I fully believe that free-market economics will solve the problem very simply, quickly and easily. Broadband ISPs must be allowed and enabled to deploy technology that will allow them to segment traffic. High quality video distribution needs to be separated out and get its own "diamond lane" delivered directly to the consumer. Likewise, the peer-to-peer traffic must be relegated to "truck lanes". The broadband ISPs can leave the rest of the best-effort traffic alone. We now have the mathematical tools and technology to describe how this traffic behaves and uses effective capacity (which I explain next).

Comcast's recent FCC challenge and lawsuit is an example of a much larger problem for broadband users. Traffic and bandwidth management (or "mediation") policies are not honestly described by broadband providers today. There may be a few reasons for this. Some conspiracy theorists would have you believe that broadband providers don't want you to know what you're buying. They don't want your average DSL and Cable Modem subscriber to actually be able to objectively compare their two offerings. There may be some truth to this. The underlying problem is really a technological one. Believe it or not, up until recently, objective measure for capacity and demand haven't easily been described and measured, at least for for broadband networks.

The Past's Technical Challenge

In 1993, a group of researchers at AT&T discovered something interesting when it comes to network traffic. They discovered that Internet traffic was actually "fractal", or self-similar in nature. This actually didn't just apply to Internet traffic, it applied to network traffic in general. For those of you in the "know", this has had some very serious consequences over the years. Specifically, it supposedly meant that ISPs couldn't easily "size their pipes". In other words, if they wanted to add another 100 users to one of their networks, they couldn't easily predict how big the pipe needed to be. Telephone networks don't have this problem. They have something known as "Erlang Formulas" to help their network operators size those pipes. The Internet has not had such a "formula". It turns out that there are multiple ways to solve this problem and they all begin by being honest with the consumer.

InterStream describes these as "mediation policies". Fundamentally, what is it that the broadband consumer buys when they pay their $30 or $50 per month for their broadband connection. This isn't objectively described in either their terms of service or acceptable usage policies today. Telephone companies didn't have this problem. Because they had Erlang formulas, public utilities commissions (PUCs) could clearly specify minimum call blocking probabilities in order to call themselves phone companies. The Internet of the past does not allow the FCC or any of the PUCs to clearly specify what it really means to be called a broadband provider.

The Present's Opportunity

It turns out that we now have the tools to clearly and objectively specify what kind of quality connection, you as a consumer, are receiving from your broadband service provider. It requires, however, that the broadband provider fully describes what kind of connection you're really buying OR that some neutral third party, like InterStream, tells you what you're getting. InterStream calls this a "mediation policy" and Internet service providers call it terms of service (ToS) and acceptable usage policies (AUPs). Broadband Service Providers get to specify what kind of service they offer while consumers benefit by being able to objectively compare different providers.

Consumers benefit through objective comparison shopping. Providers benefit through traffic and bandwidth management to meet those objective measures. Media companies can objectively specify what it really means to be a broadband TV provider. The whole industry wins by doing what they do best in their portion of the value chain.

The Future: Diamond Lanes, "Best-Effort" Lanes, Truck Lanes, and maybe a few others...



This discussion leads us to what those broadband ISPs should be doing with their networks. Obviously, the video traffic needs to get "diamond lane access" to provide a high quality experience to the consumer. Mediation policies might specify some even high quality experiences such as those for gaming or interactive video conferencing. Meanwhile, "best-effort" service and the demand for that service can now be objectively measured. In doing so, broadband ISPs and others may describe exactly what kind of service they're offering to their customers. When those highways are congested, Peer-to-Peer (P2P) applications can get the "truck lanes". Effectively, because these applications are moving very large files around and they attempt to skirt any "fair" form of behavior on the network, the P2P traffic must be shunted off to the side to make way for the premium and regular best-effort web surfers.

There are some in the ISP community that we need keep private things private. In other words, let the traffic management practices of the ISPs sit underneath the covers of their broadband subscribers. InterStream fundamentally believes that the consumers have a right to know what they're buying and the media industry should be able to deliver their content with the quality they require. It really comes down to honesty. What the consumer gets must be ultimately delivered by the ISPs. The media companies are simply using them as a conduit and they have the right to specify what kind of quality they require for delivering their content to the consumer.

Implications on Peering and Transit Internet Service Provider Agreements (to be continued...)

This leaves us with how the ISPs are going to interconnect their networks in the context of this kind of tiered Internet. I'll leave that for my next post. Until then, feel free to connect with us at info [at] interstream [dot] com.

Jeff Turner


Post new comment