D R A F T 2.4 W. B. Norton
Last Modified: 05/30/2001
Internet Service Providers and Peering
William B. Norton <wbn@equinix.com>
Abstract
Internet Service Provider (ISP) peering has
emerged as one of the most important and effective
ways for ISPs to improve the efficiency of operation.
Peering is defined as “an interconnection business
relationship whereby ISPs provide connectivity to
each others’ transit customers.” ISPs seek peering
relationships primarily for two reasons. First, peering
decreases the cost and reliance on purchased Internet
transit. As the single greatest operating expense, ISPs
seek to minimize these telecommunications costs.
Second, peering lowers inter-Autonomous System (AS)
traffic latency. By avoiding a transit provider hop in
between ISPs traffic between peering ISPs has lower
latency. So how is peering done?
This paper details the ISP peering decision-making
process. Interviews with Internet Service Providers1
have highlighted three distinct decision phases of the
peering process: Identification (Traffic Engineering
Data Collection and Analysis), Contact &
Qualification (Initial Peering Negotiation), and
Implementation Discussion (Peering Methodology).
The first phases identifies the who and the why, while
the last phase focuses on the how.
The appendix includes the description of a Peering
Simulation Game that has been used in workshops to
play out peering negotiations.
Introduction and Definitions
Internet Service Providers (ISPs) connect end-users
and businesses to the public Internet. They compete
with each other on price, performance, reliability, etc.
but they also must cooperate with each other to
provide global connectivity to all other attachments on
the Internet. The cooperation is explicitly stated at
demarcation points where they interconnect, called
imprecisely “peering points”. This is an imperfect
name since interconnection typically takes one of two
forms: a peering relationship or a transit relationship,
and both use the Border Gateway Protocol (BGP) for
routing announcement exchange. Problems arise when
the term “peering” is used interchangeably with a
“transit” relationship. To lay the groundwork for this
paper, we introduce the following working definitions.


1
Interviews with 100 ISPs over the course of two years
along with presentations of the findings to ISPs at
NANOG, RIPE and IEPG substantially validate the
findings.
Definition: Peering is the business relationship
whereby ISPs reciprocally provide to each other
connectivity to each others’ transit customers.

To illustrate peering, consider figure 1 below showing
a much simplified Internet: the Internet with only three
ISPs: WestNet, USNet, and EastNet. WestNet has
customers shown as green circles. USNet has
customers of its own (beige circles) and EastNet has
its customers shown as yellow circles.

Peering
USNetWestNet
Peering
EastNet
WestNet
Routing Table
(After Peering)
USNet
Routing Table
(After Peering)
EastNet
Routing Table
(After Peering)

Figure 1 - Peering relationships
In this example, WestNet has a peering relationship
with USNet in which USNet announces reachability of
its beige customers to WestNet, and WestNet
announces reachability to its green customers to
USNet. This is the essence of the peering relationship;
each ISP reciprocally provides access to each others
customers. EastNet also peers with USNet,
announcing its yellow customers to USNet while
USNet announces its blue customers to EastNet.

It is important to note that WestNet and EastNet can
not access each others customers in this configuration.
(The boxes below the ISPs show their respective
routing tables.) WestNet only knows how to get to
blue and green customers, and EastNet knows how to
reach only blue and yellow customers. The fact that
they both peer with USNet is inconsequential; peering
is a non-transitive relationship.

Since peering is a reciprocal non-transitive
relationship, EastNet and WestNet must peer with
every other ISP or find another way of accessing every
other ISP.

Definition: Transit is the business relationship

DRAFT Internet Service Providers and Peering W. B. Norton
2 Comments to the Author Welcome
<wbn@equinix.com>
whereby one ISP provides (usually sells) access to
all destinations in its routing table2.
Consider a slightly more complicated Internet
model below in figure 2. In this picture, EastNet
purchases transit from an upstream ISP that has
access to the entire Internet (shown as many colored
networks behind several Upstream ISPs). As a result
of this transit relationship, EastNet receives access to
all network routes in the upstream ISP’s routing table.
The upstream ISP receives and announces EastNet
routes across all of its peering and transit
interconnections. As a result, EastNet gains
connectivity to the entire Internet that is known to its
upstream ISP.
Transit $$$
EastNet
Upstream
Transit
Provider(s)
Upstream
Transit
Provider(s)
Upstream
Transit
Provider(s)
Upstream sells “Transit Services”
by announcing reachability
to the Entire Internet*
*The upstream ISP will either done announce a full routing tableor,
more commonly, announce a single “default” route for all destination
EastNet now can access the entire
Internet as seen by the Upstream ISP

Figure 2 - Transit Relationship - selling access to
entire routing table
To value the transit service, note that there are over
8800 ISPs in the US3 alone; imagine the complexity
and cost of trying to peer with all of them! In reality
you wouldn’t have to peer with all ISPs, but you
would need to have a peering relationship with them or
their upstream (transit) providers to avoid transit fees
entirely. Compared against the relatively small number
of routes received from a single peering relationship,
one can see that transit is indeed valuable and different
from peering.

2
Note that increasingly in Europe ISPs are offering and
obtaining hybrids. For example, they may purchase
“Regional Transit” from global players in a region
without adequate coverage. In a few rare cases ISPs have
arranged “paid peering” to eliminate the cost of peering
to one or the other. INSNet and GXNet for example.
3
See the Boardwatch list: http://www.thelist.com for a list
of ISPs.
As a side note, some service providers4 prefer a transit
(customer) relationship with ISPs for business reasons,
arguing that the threat of lost revenue is greater than
the threat of terminating a peering arrangement if
performance of the interconnection agreement is
inadequate.
Through the previous studies the notion of “Tier 1
ISP” repeatedly came up, and there seems to be rough
consensus on the following working definition:
Definition: A Tier 1 ISP is an ISP that has access
to the global Internet routing table but doesn’t
purchase transit from anyone.
While this distinction is discredited as being almost
impossible to prove (many ISPs claim to be a tier 1
ISP) it is an important distinction as it pertains to ISP
motivations. Tier 1 ISPs are not motivated to peer to
reduce the cost of transit; by definition, Tier 1 ISPs
don’t pay for transit.
Tier 1 ISPs may peer broadly for technical reasons.
Peering has the benefit of lower latency, better control
over routing, and may therefore lead to lower packet
loss. For ISPs that charge on Mbps, this leads to
secondary financial effects as customers use more
bandwidth5.

Recently, several Tier 1 ISPs have published their
peering policies and prerequisites on-line6:

UUNET:
http://www.uu.net/peering/

Level 3:
http://www.level3.com/us/info/network/interconnectio
n/

Genuity:

4
Conversation with Allan Leinwand, Founder of Digital
Island. Allan indic ated that the financial “teeth” are
much stronger with transit ISPs than with “peers”, and
the threat of lost revenue provides better quality and
reliability in a transit relationship.
5
This is a function of TCP: lower latency and packet loss
means that the TCP window opens more quickly and
congestion control back off algorithms are avoided. This
results in greater usage and therefore greater customer
revenue.
6
See press release:
http://www.genuity.com/announcements/news/pres
s_release_20000908-01.xml