article page | 1 | 2 | 3 | 4 | 5 |

Most of these peering arrangements are proprietary and confidential, so it is difficult to determine who runs a transit free network. But the important distinction for peering, going back to NSFnet, is that of transiting traffic; bits go in one side and out to another carrier. Peering points become very important landscape. Location. Location. Location.

Peering is designed to be a partnership of equals. What actually happens is quite different. To understand this, one needs to track the history of peering and the Internet. Around 1995, a few companies (some telco's and larger ISPs) engaged in a race to build high speed, international, IP transit networks (AT&T, MCI, BBnet, UUnet, BT, NEC, etc.). They linked their networks at points established by ICAN – the original peering

hubs. The remaining big ISPs eliminated their then significant expenses for international network terminations by merging with international telco carriers. These aggregate Tier 1 providers started laying lots of international fiber. This reduced their costs of leasing local access, and set them up to charge lesser ISPs (the Tier 2s and 3s) for IP connectivity to the rest of the world – a win-win in pre bubble times. At the height of the bubble, the big customers of the Tier 1 providers were lesser, regional ISPs. Now, post bubble, the telcos have eaten many regional ISPs who were their major customers. Thousands of other ISPs went bankrupt. Without the revenues from these smaller ISPs, being Tier 1 has become an economic burden.

Strategically today, becoming a Tier 2 provider is probably a more flexible economic position.

"Tier 2 - A network that peers with some networks, but still purchases IP transit or pays settlements to reach at least some portion of the Internet... Tier 2 networks are motivated to peer with many other Tier 2 and end-user networks. Thus a Tier 2 network with good peering is frequently much "closer" to most end users or content than a Tier 1." [Wikipedia]

Increasingly, as they interconnect their private Intranets, large multi-national enterprises begin to look and feel just like those Tier 2s. Their huge appetite for telecommunications services often translates into massive private networks with many established interconnection points to major carriers around the world. When their communications volume becomes large enough, it makes sense to establish a direct network presence at major peering points to reach outward to the World Wide Web. They do this by putting big cluster routers at or near established peering points and backhauling these over their own network fiber to their data centers or service centers. This does two things: it gets them closer (by transit lag) to their customers; and it allows them to peer with Tier 2 providers at the same location. This peering means private, negotiated cost arrangements or free exchange of traffic, with many different peers – and it means more lost access revenues for the Tier 1 telcos.

This model of Enterprise as a virtual telco has been with us for a long time. Major institutions for many years had internal telco management departments (usually staffed with ex-telco people) who understood exactly how to design infrastructure to make cost-effective use of leased lines, point to point circuits,