|Internet Interconnection :: Traffic Ratio Proxy|
In making interconnecting arrangement determinations based on balanced and imbalanced value, what is value? Seen through 'network effect,' 'value' is the benefit to one network provider of expanding connectivity to a number of end-users, in this instance through interconnection (Note that one thing 'value' is not is 'cost.' [Kenny 2013 at 10 ("In practice many commentators blur the distinction between balanced costs and balanced value, though of course the two are very different.")]). However, not all end-users are of the same ‘value.’ Adding 1,000 end-users would increase the number of people using a network and increase that service’s value. Likewise, adding one high value end-user, one for which there is great demand from end-users, might add significant value. This one end-user might be a social media star that many people want to follow, it might be high demand content that everyone wants to access, or it might be the next killer application like a game that everyone wants to play. Adding end-users increases value pursuant to network effect, but not all end-users are of equal value, making the valuation messy.
How then is value measured? You could hire a bunch of PhDs, and still quibble over whether you are getting it right. You could use the byzantine accounting method of the Bellheads, although it may be unclear whether the accounting method is correlated to value or how much the bill should be.
The key question to be answered is when is value balanced, in other words, when is it reasonable that two networks enter into a settlement-free peering arrangement. If value is imbalanced and it is reasonable for one network to pay another network for services, then, a competitive market should sort out the price. The magic question here is when is value balanced such that it would lead to settlement-free peering. And could this analysis be done efficiently such that networks can get on with focusing on their real business plans of building networks and selling Internet access?
The Internet community solved this conundrum by using a proxy. A 'proxy' is something serving as a substitute for the actual thing. The Internet community would measure relatively balanced value by employing the following proxy analysis:
- two networks are of relatively like-networks (i.e., equal size, same business plan, same geographic reach) and
- they have a balanced traffic ratio,
- then it can be assumed that the benefit of interconnection is relatively equal for the providers and it would be reasonable for them to enter into a settlement-free peering arrangement.
See History of Settlement Free Peering. [ENISA Report at 93 (“[I]n a peering relationship there is the unspoken assumption that the two ASes are in some sense roughly equal (hence the name!), so that the arrangement brings roughly equal costs and benefits to each of them. This does not mean that the peering only occurs where the ASes are strictly equal. In practice ASes will peer if they both believe that the benefit to themselves outweighs any perceived (or, indeed, actual) inequality in costs or benefits.”)] [AT&T What Does it Mean to be a “Peer”? (“Traffic ratios are one indicator of whether two networks are sufficiently similar to peer equitably with each other. A ratio that is significantly out of balance is an important signal that, in fact, the two networks are not sufficiently similar. Said differently – they are not peers.”)]
This is efficient. It is simple. It minimizes transaction costs. And it focuses on building networks and promoting access service. [AT&T Comments Dkt 10-90 2012 at 13 (“To avoid administrative overhead, parties to these bilateral peering agreements typically forgo the mutual exchange of compensation and peer on a settlement-free basis.”)] [Mair Declaration, AT&T, para. 12 ("When the ratio of traffic exchanged between the parties is roughly equal (and other criteria are met), these relationships benefit both parties by enabling them to avoid the cost of billing each other for transporting and terminating roughly equivalent traffic")] [Houston Sec. 5.2 ("Oddly enough, the parties themselves do not have to agree on what that value or dimension may be in absolute terms. Each party makes an independent assessment of the value of the interconnection, in terms of the perceived size and value of the ISP and the value of the other ISP. If both parties reach the conclusion that in their terms a net balance of value is achieved, then the interconnection is on a stable basis.")]
Notice that the proxy is a substitute and is not the same as the network effect valuation. Increases in network effect are achieved and measured by increases in the number of end-users; the proxy measures relative network size (based on things like geographic reach and presence at exchange points) and traffic exchanged (Networks set forth their geographic diversity and traffic ratio criteria for potential settlement-free peers in their peering policies. See, e.g., [IP Traffic Exchange Policy for U.S. Interconnection, LEVEL3 (2016)] [Settlement-Free Interconnection Policy for Autonomous System 2828, XO COMMUNICATIONS (2011)] [AT&T Global IP Network Peering Policy, AT&T (2016)]). These things are similar but not the same; they are close enough for the proxy to work.
The proxy is also limited to only determining settlement-free peering opportunities. Where the proxy produces a negative result, it does not determine whether an imbalance of value exists, what the actual imbalance of value is, whether someone should pay, who should pay whom, or what amount should be paid. [See Contribution by Cogent on DGCONNECT's Public Consultation 2012 at 4 ("Ratio arguments only have sense between operators of the same kind.")] Consider both an access provider and a content provider, in relationship to their upstream transit provider, are (a) not like-networks and (b) have imbalanced traffic. With an access provider, the traffic ratio is dominantly in the download direction towards the end-users. With the content provider, the traffic ratio is dominantly in the upload direction. [See NRIC Report Sec. 2.3 (discussing how the symmetry of traffic for residential Internet access providers was significantly balanced towards download, and the symmetry of traffic for content hosts was significantly towards uploads).] Traditionally both paid backbone service providers for transit service. They paid because the backbone service provided an indispensable service that they had to acquire in order to pursue their business plan. They both needed the backbone provider in order to exchange traffic with the rest of the Internet. While network effect benefited both sides of the interconnection, the benefit to the backbone service provider was insufficient to induce settlement-free peering. The terms of those transactions were determined by the market. In this case, the proxy produced a negative result both because the networks are not like-networks and because the traffic is imbalanced; but the result was negative for different reasons. The direction of the traffic imbalance and the traffic ratio does not inform who pays whom or what the price should be.
Alternatively, consider two not-like-networks, a CDN and a BIAS provider. Normally the traffic would be imbalanced in the down direction. In order to rectify BIAS provider arguments that they should be paid because of the imbalanced traffic ratio, the CDN adjusts its service so that it generates pointless upstream traffic, balancing the traffic ratio. Now the not-like-networks have a balanced traffic ratio, but still do not provide a true result for the settlement-free peering proxy. [See Hastings, Internet Tolls and the Case for Strong Net Neutrality (“But when we ask them if we too would qualify for no-fee interconnect if we changed our service to upload as much data as we download -- thus filling their upstream networks and nearly doubling our total traffic -- there is an uncomfortable silence. That's because the ISP argument isn't sensible. Big ISPs aren't paying money to services like online backup that generate more upstream than downstream traffic. Data direction, in other words, has nothing to do with costs.”).]
Where the proxy produces a positive result, where there are like-networks and balanced traffic, one can assume that there is balanced value and rationally enter into settlement-free peering. Where the proxy is invalid, one cannot make that assumption. That's all that the proxy gets you.
The settlement-free Proxy worked well in the early days of the Internet, when the delineation between backbone networks and access networks post-NSFNET was still strong - when there were different networks with very similar business plans. As those lines blurred, and providers evolved their business plans to become a bit of CDN and a bit of transit and a lot of access, one became less able to identify like-networks with balanced traffic exchanges. The Internet community has opined that the settlement-free proxy is no longer informative. [Norton, The Folly of Peering Ratios ] [Netflix Open Internet Ex Parte at 1 ("Some large ISPs continue to attempt to justify access charges based on a ratio “imbalance” between downstream and upstream traffic. But these ratios are arbitrarily set and enforced and are not reflective of how ISPs sell broadband connections and how consumers use them. Traffic volumes are consistently and significantly greater downstream than upstream and ISPs who deliver traffic over the last mile can never be in balance with the networks that deliver video. ISPs typically do not sell symmetrical Internet connections to consumers. Even though ratio-based peering does not make much sense, edge providers and ISPs still have a mutual interest in interconnecting in a way that saves costs, maximizes efficient delivery of content to end users, and makes sure that the Internet continues to scale to meet consumer needs.")] [Letter from Bopp et al at 1 (noting that the interconnection ecosystem has evolved significantly since the 1990s, stating “traffic ratios are an outdated and misleading metric for determining equality and financial burden, and are not commonly considered in ‘good faith’ discussions between content and access providers seeking interconnection”)].
© Cybertelecom ::