Peering is about trust

I have been attending the European Peering Forum in Sofia Bulgaria, and learning much. It feels as if I have inhabited the edges of the Internet- ICANN, ARIN, CIRA – until now, and finally the veil has parted and I find myself in the place where the deals are done, where the agreements are made to exchange traffic. In short, it is peering if you exchange traffic without compensation, and transit when you pay. And though it may involve money, the transaction is essentially about trust. And here is where trust is built. Knowledge of the other guy’s network is built over time by personal contacts.

This is a forum where people mentally departed from the PSTN decades ago; where telecoms, however vital, occupies layers 1 and 2, and the inhabitants of this universe are dwelling in layers 3 and up. Physical infrastructure is desirable and necessary, but it is not where the packets are exchanged.

In preparing a talk for this group, I had linguistic and conceptual difficulties to deal with. Transport of signals becomes more complex when the task is distributed across layers, as these are defined in the OSI reference model. Citing Wikipedia,

The Open Systems Interconnection model (OSI model) is a conceptual model that characterizes and standardizes the communication functions of a telecommunication or computing system without regard to their underlying internal structure and technology. Its goal is the interoperability of diverse communication systems with standard protocols. The model partitions a communication system into abstraction layers. The original version of the model defined seven layers.


Image result for osi reference model layers


 What would you call the transport of signals across the physical layer, the lowest one on the hierarchy? In my case, I would stick with the term transport. But then, in the OSI model, layer 4 is called “transport”. That is the layer at which TCP and its dumber cousin UDP do their work. Are you confused yet? The word “transport” has different connotations as one moves from carrier to Internet worlds, and until a better term is devised, the context-sensitivity will have to be endured.

The subjects of transit and peering were studied by a group called the Body of European Regulators for Electronic Communications (BEREC). Though it may read as dry as dust, BEREC published a study in December 2012 called “An assessment of IP interconnection in the context of net neutrality”. It should be mandatory reading for telecom regulators. It provides essential insights into how the Internet actually works, and how it is evolving. None of it addresses matters that telecom regulators need to regulate, on the contrary, it shows the areas where regulation would be superfluous. On the other hand it provides insights into why this market differs so profoundly from telecoms. It also makes the point as often as necessary that the separation of applications from underlying layers makes the Internet what it is, and draws attention to the need for a competitive market in physical transport to stimulate customer satisfaction and keep the ISP market competitive.

Similar points are made in Dennis Weller and Bill Woodcock’s study conducted for the OECD, called “Internet Traffic Exchange: Market Developments and Policy Challenges”, Digital Economy Papers, no.207. 

The Internet has allowed prices for connectivity to be five orders of magnitude lower than what it is for a TDM equivalent. Stated as a per minute  price for VoIP traffic the combined cost to a caller and a recipient in US dollars is 0.0000008 per minute than wholesale service providing comparable functions in TDM markets. This result has been achieved with no direct intervention of regulators.

The point here is that regulation is necessary at the level of physical infrastructure, and not at the level of transit and peering agreements. There are many reasons why competition in physical infrastructure is difficult to achieve, and why the incentives to merge into a duopoly or monopoly are strong. Thus, where competition is difficult to maintain, and players tend to consolidation, there regulators must be vigilant. By the same token, they do not need to concern themselves with a fully competitive, fluid, innovative market of the type seen here.

The matter of regulation is further complicated by the notion that players at level 3 and above in the protocol stack are supposed, somehow, to want to be carriers and to “climb the ladder of investment” and become infrastructure players. Periodically I am enabled to see clearly how inapplicable this idea is, not to mention plain stupid. At a peering forum, it is easy to see that, for the players here, the “ladder of investment” is a devolution backward, like mammals reverting to being lizards.

The moral of the story is to keep regulation vigilant where it needs to be, and absent where the market is working.

To top