The Internet Connectivity Market

Published Date: 02 Nov 2017

The economic puzzle of the Internet connectivity market

1.1 Introduction

The Internet is a collection of thousands of private networks that interconnect and create the "network of networks" - The Internet connectivity market is the fundamental economic setting that enables business transactions among companies of different size and scope that own private networks and want to interconnect. It is a dynamic, unregulated environment, where companies with different market power and culture that compete for market share, have to collaborate to achieve global connectivity.

The recent Internet growth in terms of both demand, through thousands of new customers that use the communication infrastructure globally, and supply, through the introduction of new sophisticated services and applications, sets new challenges for the Internet con­ nectivity market and exerts strong pressure in the p,xisting structure. A key challenge involves provision of network service quality (e.g. performance guarantees in network ser­ vices), which is a prerequisite for the deployment of many value-added applications (e.g. real time video, video conferencing) and consequently their commercial exploitation. This thesis attempts to make a modest contribution both to our understanding of the market dynamics as well as to the key market players by proposing the introduction of new pricing

1

schemes that may induce net :vork service quality provision in interconnection agreements.

The Internet connectivity market has no traditional market characteristics. It has evolved along the growth of the Internet. The Internet was initially developed in the United States by public initiatives, co-operation, peaceful co-existence among key players, and lack of commercial interests. At that time, connectivity to the backhone network, the NSFNET, was offered free of charge, for enabling knowledge transfer among researchers. When Internet Service Providers (ISPs) came into play to provide services to the business community, Internet Backbone Providers (IBPs) emerged to offer connectivity to the global Internet community. This was accomplished by interconnecting the H3Ps' networks with each other and with the NSFNET. At this stage researchers' interests were focused on the role of compatibility as means of exploiting network externalities [36], [39].

Private sector involvement rapidly prevailed over public and led to the development of the current Internet. This development offered new directions to researchers and practitioners to set economic principles in the Internet connectivity market. Initially, interconnection was considered a commodity and a necessary service to achieve global connectivity. l\fany practitioners implicitly assumed that such service would he financed by public initiatives due to its obvioUH value. Thus, they argued that it should he offered free of charge. The emergence of private companies, such as the IDPs, with core business activity the sale of connectivity services, underlined the role of pricing. However, interconnection agreements between the private networks were developed ad-hoc and remained that way even after the huge increase in provision of Internet access services.

Interconnection agreements have direct implications on the Internet access services offered to end-users/c11Storners, which are not comprehensive end-to-end services. vVhen traffic passes from an ISP's customer to a customer of another ISP, the sender ISP may pa.y for the traffic that passes to the next ISP (e.g. transit). When the traffic leaves the sender ISP's "accessed" networks and passes to another network (e.g. receiver's ISP) the sender ISP ceases to pay the further traffic's transport. At that point the receiver ISP implicitly

2

assumes responsibility for payment of the traffic's transport for the second part of the transmission path [67J.

Even if service quality is guaranteed within a network, the actual performance depends on conditions of several networks that mediate packets' transport throughout the world. In

1tl1is context the sender cannot rnonitor

or control, tl1e network services offered to its traffic

by the various intermediary ISPs that affect network service quality experienced by the receiver. Thus, the sender experiences an asyrnmetric information problem (i.e. it cloes not know the action taken by intermediary ISPs to serve its traffic when passing through their networks) that affects the service outcome. In particular, each intermediary ISP will take the least cost action to serve this traffic even if such action may have detrimental effects for the sender's benefit. The lack of direct payments for end-to-end packets transport impedes provision of network service quality because the ISP has no economic incentives to offer more t}1an "best cffort1' services.

The basic problem faced in the Internet connectivity market is to ensnre that each ISP is fairly compensated when its network resources are used. In case of low traffic load on a network, serving a new customer 1s traffic does not generate high marginal cost. However, during periods of high traffic, or congcstion 1 efficient resource allocation becomes a great concern and existing customers experience negative externalities from the extra traffic of the newcomer. In this case there is high marginal cost for serving the newcomer. This cost can be described as the opportunity cost of reducing the service quality offered to other customers (e.g. higher average delay). The negative network externalities become more intense when customers have specific service quality requirements for the end-to-end service.

Nevertheless, pricing of interconnection service is not straightforward because its value cannot be quantified explicitly. For example, end-users have an implicit expectation of universal connectivity whenever they log on to the Internet, regardless of which ISP they choose. The end-user does not value connectivity as such. Instead, an end-user expects a

bundle of services that include global access, high performance and universal connectivity.

The adversaries of pricing interconnection service argue for bandwidth over-provisioning. They believe that instead of using complicated pricing schemes it will be better to increase network capacity in specific locations that experience high traffic loads [101]. However, without a coordinated action of all the involved networks this solution cannot work. Co­ ordination is difficult in this market bL'Cause of the large number of involved parties with different strategics that arc spread around the world. Moreover, there is no intense compe­ tition that may induce every participant to proceed in such investment decision. Further­ more, bankruptcy of some Il3Ps has been partially attributed to the high cost involved in overprovision of bandwidth (104].

The adoption of new pricing schemes for interconnection agreements that allow for cost compensation among ISPs may sustain a diverse and well-populated market. These schemes should offer incentives to the ISP to provide more than "best effort" services. [n 1993 MacKie-Mason and Varian [91J noted, "we think that the major network service challenge in the next two years will be to find ways to support network interconnection .... We think it is inevitable that system of settlements will emerge."

The purpose of this chapter is to trace the dynamics of the Internet connectivity market and generate insight on how the research questions should be formed. The following section underlines the key economic and technological characteristics of this market and presents the rnsearch's motivations. Subsequently, the research objectives are stated. The chapter concludes by recapitulating the ma.in characteristics of the Internet connectivity market and by illustrating the outline of this thesis.

Motivations for research in pricing on the Internet connectivity market

The Internet connectivity market structure

The Internet market is structured hierarchically, comprising three main levels of partici­ pants, namely, end-users, ISPs and 113Ps. End-users are at the bottom of the hierarchy and access the Internet via ISPs. End users include individuals and business customers. At the top of the hierarchy, 113Ps own high speed and high capacity networks to provide global access aJld connectivity. They sell primarily whohsale Internet connectivity services to the ISPs. ISPs then, resell connectivity services, or add value and sell new services to their customers. However, IBPs may also get involved in ISP business activities by selling retail Internet connectivity services to end-users. Both IBPs and ISPs provide complementary inputs to the bundled network services that end-users consume (see Figure LI).

[Ds;i

-

--·•.. - ·/•••1Ir

User

::"

:;;; ;d;,

User

Figure Ll: The Internet connectivity market's structure

User

The hierarchical market structure of Internet connectivity involves two main types of pric­ ing contracts: pricing between end-users and lSPs for primary Internet access and pricing between ISPs and Il3Ps for interconnection. Thus, two markets for Internet connectivity

are identified. retail and wholesale for global access and connectivity to end-users, respec­ tively. This thesis focuses on the wholesale market.

For the purposes of this thesis, the Internet connectivity market's key players have been grouped as follows:

Internet Backbone Providers (!13Ps) (e.g. [27]) are owners and managers of core IP networks, which have wide (aln1ost global) coverage. They develop, n1aintain and upgrade the global network infrastructure. Il3Ps play the role of network service "vvholesalers' . IBl")s offer global access to business customers and interconnection services to national or local ISPs through their high capacity networks. In order to offer global access and connectivity they mainly establish peering agreements with each other. The backbone market is an oligopoly, dominated by few global players

(e.g. MCI-Worldcom-UUNET, SPRINT). !13Ps mainly offer flat-fee and usage based

pricing scheme,;; especially for business customers.

Transit ISPs (e.g. [65], [72], [45]) mediate between two networks that do not in­ terconnect directly. They buy band width capacity from Il3Ps and sell connectivity service to ISPs. Transit ISPs are intermediaries between "wholesalers" and "retail­ ers" of network services and facilitate connectivity of local/regional networks with global ones. In many cases, transit ISPs (e.g. AboveNet/MFN) are not involved in any kind of service operation, or service maintenance, they simply resell connectivity service without adding any value. They also compete with ISPs for access services provided to business customers. They mainly charge flat fee, or usage-based prices. Some of them (e.g. Abovenet/MFN, Verio, Teleglobe) offer transit service free of charge in order to increase their market share.

Internet Exchange Operators (e.g. [6), [99]) manage and operate the physical infras­ tructure for Internet traffic exchange, the Internet exchange points (IXPs) or Network Access Points (NAPs). Those are multiple access networks that support the poten­ tial for any to any communication in order to localize traffic exchange. IXPs are

6

dedicated to transit or peering, or have a combined, hybrid approach (e.g. lnterNap) [99]. The !XP business models are mainly based on flat-fee pricing structures tha.t do not correspond to the lSP's traffic volume being exchanged, which is not metered, given that access facilities a.re leased by the participating ISPs [19].

Internet Service Providers (ISPs) (e.g. [48], [52]) cover the "last mile" between end­ user/custon1er and IBP, using copper or fiber lines, or even wireless technologies, from telecommunication companies. They play the role of network service retailers, providing connectivity services and local Internet access, through packet switched networks. They build, or lease physical network infrastructure and buy transit ser­ vices in order to have global access and connectivity. At the national and the local level, they estahlish peering agreements. The ISPs' market is a highly competitive not regulated environment [35]. Recently, customers' requirements for network service quality enabled them to diversify their services and develop new business activities

(i.e. \Neb hosting, Virtual Private Network services) [72]. ISPs offer flat price and usage-based contracts, or Service Level Agreements (SLAs) to their customers.

Customers can be classified according to their propensity to consume network ser­ vices in two broad categories; businesses and individuals. Bu.siness Cu.9tomers (e.g. [21]) include every business entity that uses network services for generating revenue. For example, content providers are business customers that provide information and content related services to end-users. They generate revenue from selling content and expect high network performance. Their requirements for network service qual­ ity may be described in an SLA. Individual Cu.stomers (e.g. [21]) are people who "consume" network services for personal use and mainly have unpredictable usage patterns. Individuals have lower propensity to spend money for network services than btrniness customers. This difference is critical when considering cost allocation and pricing strategies for network services.

The key players, IBPs, ISPs and Internet Exchange Operators, form interconnection agree-

rnents that enable them to offer global access and connectivity to their customers. The interconnection agreement includes a settlement on whether the traffic exchanged vvill be charged. The Internet connectivity market operates with two generic types of interconnec­ tion agreements; peering, the exchange of traffic in a settlement-free basis and transit, a customer-provider relationship mostly based on flat price settlements. The fundamental characteristics BJld linlltations of interconnection agreen1ents that 1notivated research of this thesis are presented in the following sections.

Peering

A peering agreement involves the exchange of traffic and routing information bet\veen t\vo net rorks at no interconnection charge. Thus1 traffic is effectively exchanged bet\veen networks under a barter arrangement. This exchange may occur publicly at IXPs and NAPs, or privately between two networks. The agreement is non-transitive, which means that ISPs are not exchanging customer routes learned through peering with other ISPs. Peering partners only exchange traffic that originates from the customer of one partner and terminates to the customer of the other partner, without providing transit to the peering partner's traffic on another peering partner. Peering costs include the purchase of equipment and the provision of network capacity needed (e.g. router ports and links). Peering partners mainly exchange their customers' traffic at the geographically nearest possible exchange point and provide "best effort" services when terminating traffic. Four forms of peering agreement are identified with respective benefits and challenges (see Table 1.1):