The core of the global Internet operates informally, almost completely without written agreements.
The Internet is a Big Deal. It’s at least as significant for mankind as was the advent of the printing press (in the 15th century) or writing (several thousand years BCE). Like writing and printing, the Internet is globally pervasive (it’s even in North Korea – to some extent). Yet a recent survey conducted by Packet Clearing House shows the commercial arrangements on which the Internet depends are almost entirely informal, without written contracts. At first blush, that’s surprising.
Packet Clearing House (PCH) analyzed 142,210 Internet carrier interconnection agreements (a.k.a. peering agreements) between 4,331 different ISPs representing approximately 86% of the world’s Internet carriers (i.e. 86% those ISPs whose networks form the global Internet backbone). These organizations are incorporated in 96 different countries. Of the 142,210 agreements, only 698 (0.49%) were formalized with written contracts. Again, these agreements pertain to the Internet backbone, not to access networks (where political issues reign and written contracts are widespread).
How does it work? The Internet backbone consists of a little over 5000 organizations that have agreed to exchange traffic for their mutual benefit. Individual organizations may exchange traffic with just a few others or with many others but the process is routine. There are well established standards like Internet Protocol (IP) for packets and Border Gateway Protocol (BGP) for routing, an extensive shared vocabulary and a set of common understandings, e.g. only packets for customer networks are exchanged and each network exercises reasonable care to prevent abusive or criminal misuse of the network. Because there are many potential alternate paths to reach a given destination IP address, specific peering arrangements can come and go at the discretion of either party. Loss of a specific peering connection doesn’t (typically) result in loss of connectivity, just a change in the cost of carrying some traffic.
Finally, in almost all cases there is no exchange of money, only an exchange of traffic. For many organizations, this likely reduces the need to involve their lawyers.
In any event, it’s nice to see PCH quantify something I’ve suspected for many years (since reading Gordon Cook’s 2002 article which showed how “donut” peering was breaking the original backbone oligopoly).