National Research and Education Networks: Analysis of Management Issues

Dmitry GALAGAN <d.galagan@is.twi.tudelft.nl>
Maarten LOOIJEN <m.looijen@is.twi.tudelft.nl>
Delft University of Technology
Netherlands

Abstract

This paper addresses the problem of creating an effective conceptual model of management of National Research and Education Networks (NRENs) that reflects new dynamic changes in the whole Internet management structure. The analysis includes the classification and evaluation of important management issues, the comparison between various European NRENs, and main trends in the management of NRENs.

The paper attempts to give a picture of research and education networking, which could facilitate and promote further development of NRENs and the Internet. The analysis presented in the paper can be a source of information for managers preparing recommendations for the improvement of NREN management and establishing new services for target user groups.

The work presented in this paper forms part of an ongoing doctoral (Ph.D.) research project carried out at Delft University of Technology.

Contents

• 1. Introduction
• 2. Users of NRENs
  • 2.1. Commercials: to serve or not to serve?
  • 2.2. Classification of users
  • 2.3. End-users community
• 3. Services provided by NRENs
• 4. Organizational structure of NRENs
  • 4.1. The role of the backbone operator
  • 4.2. Hierarchical models of NRENs
• 5. NREN funding models
  • 5.1. NREN costs and income
  • 5.2. The model of money flows
  • 5.3. Funding models
  • 5.4. Pricing schemas for transmission services
• 6. Summary
• 7. Acknowledgments
• 8. References
• 9. Author information

1. Introduction

National Research and Education Networks (NRENs) in Eastern European, the former Soviet Union, and developing countries are being developed very fast under the conditions of evolving information technologies. NRENs are becoming an important part of National Information Infrastructures and a place where new technologies are tested and implemented. The processes of building new networks and developing existing ones demands both technical skills and management knowledge (organizational, economical, financial, policy, and legal aspects). The importance of management aspects has been demonstrated by the success of the recent NATO/TERENA/CEENet workshop "Management and Policy Issues in running a NREN" (see [1]).

Current information about management of NRENs is presented by different operational and organizational models, usually in the form of a simple description of particular experience and existing practice. Because of this, management is often carried out by the trial-and-error way, which often leads to a waste of money, time, and labor.

The solution that could allow avoiding such a waste of valuable resources is a conceptual model of NREN management. The creation of such a model is topical and important for the further development of NRENs and the Internet. This model should reflect and classify the best existing practice and experience. Its main goals should be to advise how to improve the management of existing NRENs, as well as to implement appropriate management models for building NRENs (an example of building an NREN is the Ukrainian Research and Academic Network; a project started in autumn 1998).

The development of such a model requires detailed data about existing NRENs and their management. The authors have collected such data. The data about management issues have been collected from various sources: literature, proceedings of networking conferences/workshops (INET, JNC, TNC), Web sites of European NRENs (see [9]), and Japanese SINET (see [10]). These were of particular importance for the preparation of this paper [1]. All the collected "raw data" can be found here.

The collected data were analyzed by classifying, estimating, and comparing different management issues. The management issues are considered in the next four sections of the paper: Users, Services, Organization,and Funding.

The paper in the framework of the research project

The paper presents an initial part of the ongoing research carried out at the Department of Information Strategy and IT Management, Delft University of Technology in the frame of the doctorate (Ph.D.) research project "Modeling Management of Research, Education and Academic Networks" (see the project's home page). The research project began in March 1998 and should be finished in April 2002.

The general aim of the research project is to get a better understanding of research and education networks and their management by means of developing a model that supports such a management. The research questions are raised for fulfilling this aim. The one relevant to this paper is "What aspects play a key role in the management of REA-networks?". More information about the research and research questions can be found in the Research Proposal.

A model of management of research and education networks, which should be created during the research, will be based on the general theoretical approach to management, control, and maintenance (MCM) of information systems developed in Delft University of Technology. The main building blocks of this approach are the MCM paradigm, the extended state model, and the triple model of management. According to the mentioned approach, a set of management tasks is formulated, which is further divided into task fields and task areas. All management tasks are viewed on three levels: strategic, tactical, and operational. Further, MCM of information systems is divided into three forms: technical, application, and functional management. This approach has shown its usefulness and applicability in a number of practical cases. More information about it can be found in [2].

2. Users of NRENs

2.1. Commercials: to serve or not to serve?

Because the primary aim of NRENs is towards servicing research and education institutions, which are usually noncommercial, there is a challenge for NRENs as to whether to serve commercial organizations or not. Most European NRENs do not have commercial users. Some NRENs have users that belong to commercial research and education (private universities, colleges, etc.). For example, commercial research organizations can have access to GARR in the frameworks of agreements or projects with research and education institutions, which are the regular users of GARR. A few NRENs have commercial users from industry (JANET, NASK, and HUNGARNET).

The acceptance of commercial users seems to be dependent on the funding model of an NREN (see details of the financial models in the section Funding). If an NREN is subsidized by the government it is not likely that commercial users are served. If an NREN is user-financed, usually an association of users decides whether to serve commercials, and in some cases this decision is in favor of serving commercials. In some countries the decision is made on a regional level, e.g. a regional network can decide whether to serve commercials (JANET, RENATER).

2.2. Classification of users

The collected data have allowed us to make the following classification of NREN users (see figure 1). Considering the information given in the previous subsection, the first characteristic of the proposed classification is the profit status of a user. In this way all users can be divided in two groups: noncommercial (public) and commercial. Further, users from both groups are divided according to their belonging to either the research or education sector.

There are also users that don't belong to either the research or the education sector, such as libraries, public authorities, ministries, museums, and industry. They form separate user groups. Libraries play a particular importance for NRENs. Due to the fact that libraries are natural "warehouses of information," they can provide the research and education community with the necessary information, potentially being the best content providers.

The occurrence of each particular user group is growing from the bottom of the figure to the top. Thus, public universities are the most widespread users, while general commercial bodies (industry, small and medium-sized enterprises) are the least occurring users. However, this rating is not strict, has no exact quantitative or statistical ground, and is based more on the general impression of the authors.

Figure 1. Classification of NREN users

The trend, which can be noticed by studying the history of NRENs, is the expanding of the target user base. Many NRENs at the time of their creation were intended to serve a specific set of users (for example, only universities or academy of sciences). In some countries there is still a separate network for research and a separate network for education. Examples are the Greek university network GUNET and research network GRNET, and the Danish research and academic network DARENET and education network SEKTORNET. But over some time, new categories of users were accepted and are still being accepted, including commercial users. Nowadays it is becoming more common for NRENs to serve organizations that do not belong to the research and education community. John Dyer has noted in [3] as a common theme for many NRENs cross sector partnership and collaboration, which usually requires good communication infrastructure between collaborating sectors.

2.3. End-users community

NRENs and their services are directly or indirectly used by people (end-users). End-users can be divided into two categories: regular users and administrative users. Administrative users are those people involved in the process of campus networks' management (technicians, system administrators, etc.); they are the networking professionals.

Among regular users there can be distinguished several categories: students (scholars), teachers (academic staff), researchers, and others. Further, researchers can be divided into researchers-networkers, doing research related to networks; and researchers-non-networkers, using network as a product within their research.

Each category of end-users has various requirements of the network and its services. End-users belonging to different categories have different knowledge and experience with networks, and use networks for various purposes.

The purposes of using the network and activities, which are engaged upon utilization of the network and network services, seem to be important issues for many NRENs. In particular it concerns the use of bandwidth and generating traffic streams, especially through expensive international links. For example, the Joint Information Systems Committee (UK, JANET) has recently commissioned a study on the reasons for network use (see [3]).

3. Services provided by NRENs

The most comprehensive and structured description of NREN services was found on the JANET Web site. In [4], all services are classified into operational, informational, security, and support services. This document also gives a clear definition of service levels and performance indicators for each particular service or service group by means of time, numbers, and percents (availability, mean time between failures, end-to-end latency time, queries, processing rates, etc.). During the investigation, no other documents were found with such a clear definition of service levels.

The JANET classification, together with some changes and additions (to make it more universal and suitable for other NRENs), was the basis for the classification of services, presented in table 2.

* 1 - most common, 2 - rather common, 3 - least common

Each service has a different occurrence among the sets of services offered by NRENs. Some services, such as basic transmission services, Usenet news and Web/ftp caching, are offered by all NRENs. But some services, such as intermittent connectivity (dialup) and many of the user support services, are offered only by a few NRENs. The collected data have allowed an estimate of services' occurrence and to place them in three groups: 1 - most common services, 2 - rather common, 3 - least common services (see the last column of the table 1).

"External networks access provision and transmission" are the most used and most expensive services (especially access to the Internet). International links are a major area of traffic growth. For example, 50% of traffic, generated by the users of JANET, is destined for the United States (see [3]). Transatlantic links of most NRENs often become congested.

Most NRENs have connections to one (less often, two or more) Internet eXchange (IX). Some operators of NRENs are also operators of IXs (ACONET - Vienna IX, SURFNET - Amsterdam IX /till 1997/, HEANET - Irish Neutral IX, GRNET - Athena IX). Having a connection to an IX seems to be very important for the national connectivity, e.g. possibility to peer with networks operated by commercial Internet service providers (ISPs) in a country. This prevents long routes, when traffic between two hosts in the same European city travels through the United States and the saturated transatlantic links. Both parties (NREN and ISP) usually benefit from such a peering agreement. For more information about European Internet Exchanges see [8].

Dialup service is not usually provided by NRENs, it is supposed to be provided by campus networks. The exceptional case is SUNET, which offers cheap dialup access for students (~2.5Ecu/month) by outsourcing this service to a commercial ISP.

Messaging (e-mail) services usually are not meant for providing mailboxes for end-users, this is supposed to be done by campus networks. These services are rather meant for the relaying of messages (in case of a temporary unavailability of a connection to a destination campus mail-server) and for converting messages between different mailing systems (such as the gateway between SMTP and X.400).

Network monitoring/accounting services are continuously gathering traffic information, measuring load status of links (especially international/transatlantic links), and presenting all this information in various kinds of graphical reports and statistics, which usually can be accessible online (via Web-interface). This service is especially important when the use-based pricing schema is used rather than the flat rate one (see details of these schemas in the section Funding). UKERNA (JANET) charges its clients for the use of transatlantic links on the basis of the transmitted amount of data. Regular users do not usually need this service, rather it is very important for administrative users.

Not many NRENs offer Web-based information services, which are also called "content provision." The authors could find only a few examples of such services:

• HUNGARNET Electronic Library project and Common Electronic Catalog project. The goal of the first project is to make electronic documents available by collecting, organizing, and transforming them to a uniform format using "human labor." The documents can be accessed through various interfaces, such as gopher, ftp, www, and e-mail. The second project aims at implementing the distributed search in the various integrated library systems of Hungary.
• SURFNET InfoServices (in cooperation with Royal Dutch Library), a part of it is NL-menu service -- a structured list of Dutch Web-servers with a powerful search engine.
• SINET (Japan) information services. The purpose of the Career Information Service is to gather and provide information on and advertise posts in Japanese universities. Catalog Information Service is the online cataloging system with the main purpose of constructing the union catalogs of books and serials (GIF files) covering the whole country through cooperative data entry (shared cataloging) from participating university libraries. Electronic Library Service provides an integrated system of bibliographic databases and an electronic document delivery of Japanese academic journals on the Internet. In the Information Retrieval Service, more than 73 million records of scientific information in the fields of humanities, social sciences, and natural sciences have been accumulated for online supply to researchers.

The last example (see [10]), being not a European one, seems however to be a unique one, worth mentioning here. The authors could not find analogous information services among European NRENs. The mentioned services are particularly aimed at the research and education community and its information needs. It is believed that such a provision on national and international levels is the future of information provision in a global information infrastructure and it should be implemented in other countries and their NRENs. The development of national and European information content also encourages the use of cheaper national and trans-European links (TEN-155), rather than congested expensive transatlantic links.

Security services include setting up firewalls and access-lists (packets filtering), and a CERT service (Computer Emergency Response Team). CERT activities include monitoring of security incidents (for example, SPAM attacks), dissemination of information about such incidents by means of sending mail and news messages to users, organizing training courses/seminars, and other activities.

Customer service (helpdesk) is usually not a direct service for regular end-users, but for administrative users. The latest are sometimes called user-contact persons (one or two persons per each connected institution).

Some of the user support services, which belong to the training services, are offered by international organization such as TERENA and CEENet. These are the international organizations devoted especially to research and education networking.

4. Organizational structure of NRENs

Many research and education communities in Europe often use the following organizational structure. Association of Users of an NREN is established in a country. This Association is usually founded by one or more government bodies (ministries of science, education, etc.) and a number of research, education, and academic institutions -- the primary users of an NREN. Some NRENs allow commercial users to be members of such an association, too. For example, HUNGARNET has two categories of membership: ordinary membership (research and education institutions) and sponsoring membership (any Hungarian or foreign person or organization).

Each member of the Association delegates representatives to the Assembly (meeting of users). Assemblies take place on either a regular (once or twice per year) or ad hoc basis. Such a meeting allows getting feedback from users, discussing important issues, and making strategic decisions concerning management of an NREN (funding, investments, policies, etc.). Assembly usually elects a management board and a technical advisory body. The management board makes strategic decisions between consecutive meetings of the assembly. The technical advisory body is meant to advise the management board, assembly, and the users on technical (networking) matters. Usually, networking experts from participating institutions form this advisory body.

In the table below some examples of the bodies mentioned above are given.

4.1. The role of the backbone operator

The next important element in the organizational structure of an NREN is the backbone operator. The backbone operator is directly associated with an NREN, it is the main body, responsible for actual day-to-day management of an NREN. This management is also called the "operation" or "running" of a backbone. Most of services, described earlier in the section Services, are provided by the backbone operator. Especially it concerns operational services (see table 1). However, basic transmission services on a physical level (ATM, FDDI, leased lines, satellite channels, radio-links connections, etc.) are usually operated by a corresponding physical infrastructure provider (PNO, satellite provider, radio-links provider, etc.).

Sometimes the backbone operator is the computer center of a technical university. In some cases it is a separate legal entity (see examples in table 2). There are also a few examples when the backbone operator is the division of a government body such as the Academy of Science or Ministry of Research.

Most backbone operators of European NRENs are not-for-profit organizations. John Dyer has shown in [3] that provision and coordination of national networking services by the not-for-profit or government-run organization (UKERNA) is the right approach for the United Kingdom and for many other countries. He has referred to the experience of the Internet2 project in the United States and the formation of UCAID -- the not-for-profit association for the coordination of the advanced Internet development.

The exception among European NRENs is SURFNET, which is operated by the for-profit company.

4.2. Hierarchical models of NRENs

In addition to the backbone operator, there could be regional operators in the generic organizational structure of an NREN. This is related to the hierarchical organizational models of NRENs. There are two major hierarchical models of NRENs: a two-level model and a three-level model (see figure 2). Each level represents not only the technical network infrastructure (such as links, routers, etc.), but also the management structure and the relationships between the users and the service providers.

In the two-level model, campus networks are connected directly to the backbone, more precisely to the backbone points-of-presence (PoP) (e. g., to the routers operated by the backbone operator). SURFNET, ACONET, and BELNET use the two-level model.

Figure 2. Two hierarchical organizational models of NREN

In the three-level model there is an intermediate level -- regional networks. Campus networks are connected to a regional PoP (e.g., to a router operated by a regional operator). Each regional operator is connected to a backbone PoP. A regional network is sometimes called a metropolitan network, if it covers an area of metropolis. DFN, RENATER, and JANET use the three-level model.

In addition to these three levels there could be the fourth one -- the international level. There are two international research and education networks in Europe: TEN-155 and NORDUNET. Five Nordic NRENs (FUNET, SUNET, ISNET, UNINETT, and DARENET) are joined into NORDUNET for the common utilization of the external channel (318 Mbps, February 1999). This case is different from the TEN-155 network, which is the international research and education network too, but each participant has its own external channel to the Internet.

John Dyer has noted in [3], as the common issue for some European countries, the process of regionalisation (e.g., the development of regional networks). The main incentives for this process are better local support, better collaboration, and lower costs due to the smaller distances involved, in comparison with the two-level hierarchy. Regional management is able to gain a better understanding of the local competitive position with respect to the service suppliers, and hence archive better deals in a service provision.

5. NREN funding models

5.1. NREN costs and income

For any business activity, including running of a network, the following two aspects play important roles. These are the costs (expenses) and income.

Two groups of network costs can be distinguished:

• development costs: upgrades of backbone and external channels' bandwidth, equipment upgrades, innovative networking projects, etc.;
• operating costs: costs of connections, service costs, personnel costs, the depreciation of equipment, administrative expenses, etc.

For covering of the network costs, income should be available. The sources of income can be divided in the following categories: user fees, government subsidies, and donor grants. User fees usually do not cover development costs but operating costs. This is a periodic source of income, which usually comes monthly or annually. Government subsidies can cover both operating and development costs. Examples of government bodies, responsible for the allocation of subsidies, are Ministries of Science, Technology, or Education. Donor grants usually cover development costs. Sometimes grants cover operating costs as well, in particular it concerns networking projects in the former Soviet Union and developing countries. For such projects during a limited period of time, operating costs are partly or entirely covered by the external grants allowing an NREN to develop and grow.

The well-known donors for networking projects are Soros Foundation (Tempus Tacis Programme), NATO (Scientific Panel), and the European Community (Esprit Programme, Domain #6 - High Performance Computing and Networking). A grant usually comes as the result of a competition among different proposals. Therefore, special attention should be given to the rules and practice of making such proposals, because the funding mainly depends on the success of such a proposal.

5.2. The model of money flows

The costs and income should be viewed in relation to the legal entity, which represents an NREN. This is the backbone operator, responsible for the running of an NREN. All means that are spent (costs) and obtained (income), usually go through the backbone operator. For the visual purpose of the coming analysis, the model of money flows is proposed (see figure 3). Boxes represent different legal entities participating in the financial life of an NREN. Arrows represent money flows. The arrows going from the box 'Backbone operator' represent outsourced parts of the costs, e.g., means that are paid to other organizations (the upper boxes on the figure). The costs that belong to the backbone operator's expenses (such as personnel costs, administrative expenses, etc.) are not shown on the figure. The arrows going to the box 'Backbone operator' represent the sources of income.

The presented model allows examining not only financial relationships among various legal entities, but also their functions and responsibilities, thus giving more insight in the organizational aspects of the NREN management.

Figure 3. The model of money flows

PNO is a Public Network Operator, e.g., a national telecom operator. In many European countries it is still a single company, which exclusively provides terrestrial transmission services inside and outside of a country. This monopolistic situation began to change a few years ago, nevertheless, in practice the competition in this area does not yet exist in many European countries or is only at in embryo state.

The Internet Service Provider (ISP) is the main provider of the service "Access to the Internet." Usually this is one of the big American ISPs, such as MCI, Sprint, Teleglobe, or Unisource, which form the backbone of the Internet. It could also be a European ISP, such as Ebone or EuropaNET. As has been already noted in the section Services, most NRENs peer with ISPs via Internet eXchanges. The peering agreements are usually free of charge and to the benefit of both contractors -- NREN and ISP.

The subcontractor is an organization that provides the services that are not provided by the backbone operator, ISP, or PNO. It could be a regional operator or a provider of transmission services (such as satellite or radio-links provider). In the case of SURFNET, subcontractors are the computer centers of four Dutch universities and some other organizations. The operator of TEN-155 (DANTE) is another example of such a subcontractor.

The difference between the costs and the income is the profit. Most backbone operators are not-for-profit businesses. The exceptional case is SURFNET, which is operated by the commercial company (SURFnet ltd.). This operator was the only one whose financial records were found during the data collection process (browsing of the Web sites). All other NRENs do not publish their financial records.

5.3. Funding models

According to the ratio between user fees and government subsidies in the operating costs (usually only this part of the costs is considered), it is commonly accepted to distinguish two major funding models for NRENs: the user-financed model and the government-subsidized model. However, because of the fact that many European NRENs have both sources of income -- user fees and government subsidies -- the mixed model can be distinguished as well. The examples of European NRENs using these models are given in table 3. BELNET, SUNET, and SWITCH can be ranked to the mixed-financed NRENs because some users are subsidized, while others are 100% charged for the network services.

Notes:
(1) The percentages were obtained from [1].
(2) Universities and some other categories of users are partially/entirely government subsidized.

User fees reflect services that service providers provide to users. The backbone operator (the main service provider) charges users for the services it provides itself and for the services that are outsourced to other service providers. Some services are offered for free. These could be, for example, Web/ftp caching and mirroring, e-mail/messaging services, and Usenet news service. Other services are charged either periodically (monthly or annually) or non-periodically (for services that are used occasionally, such as user support services).

The biggest part of a user fee is the fee for transmission services, e.g., first two groups of services, presented in table 1. This fee covers the operating costs of the backbone and the international links. International links are always a big concern for NRENs. They are always saturated, being the weakest point in NREN infrastructures. It especially concerns Europe - U.S. transatlantic links. Extremely high prices on international/transatlantic links in Europe (compared with the United States) are mostly the result of the recent telecommunications monopoly.

5.4. Pricing schemas for transmission services

There are two major pricing schemas (models) according to which NREN clients are charged for transmission services. These are the flat rate and the usage-based pricing schemas. The overwhelming majority of NRENs use the flat rate schema. Each fixed-bandwidth connection is charged a fee, which allows unlimited use up to the physical maximum flow rate (the bandwidth of connection).

There are a few NRENs that use the usage-based pricing schema for their transatlantic links. The basis for this pricing schema is the amount of data transmitted by a user through a (transatlantic) link. The only European NREN that uses this schema is JANET. Usage-based schemas are also used by NRENs in Chile (see [6]) and New Zealand (see [7]).

The usage-based pricing schema has some disadvantages in comparison with the flat rate one, as well as some advantages. The main disadvantage of the flat rate schema is the high cost of the accounting/charging system that includes costs of personnel, hardware/software costs, and administrative expenses. Routing equipment, which needs to be installed and/or upgraded, must have a better performance to cope with the traffic accounting tasks. The next important disadvantage of this schema seems to be important for the government-financed organizations in particular. These organizations receive periodically (often once a year) fixed funds, which are usually planned in advance. Because network traffic can hardly be planned, there could be a fear from the users' side about the under- or over-estimation of funds that should be reserved.

The advantage of the usage-based schema is that it is possibile to move to a variable quality of service model, which is gradually being implemented on the Internet (traffic prioritization, bandwidth reservation, etc.). The comprehensive statistical data, which are coupled with the accounting information, allow optimizing the use of network resources. Being implemented on a campus level as well, the usage-based schema can encourage end-users to be more careful with the use of critical network resources, such as external channels. In this way, the use of the network for non-legitimate purposes (such as leisure) can be reduced.

For more detailed information about the pricing schemas refer to [5].

6. Summary

Below a short summary of the paper is given with some major results of the conducted analysis.

The classification of users was given in the section Users together with the estimation of the occurrence of each particular user group. The issue of servicing commercial users was considered. Some issues related to end-users were considered.

The classification of services together with short descriptions of some services was given in the section Services. Special attention was given to information services (content provision).

The generic organizational structure of NRENs was described and analyzed in the section Organization. The relationships among various bodies forming this structure (association of users, management board, technical advisory body, and backbone operator) are analyzed. The various hierarchical models of NRENs were analyzed.

An analysis of the NREN costs and income was presented in the section Funding. The proposed model of money flows gives more insight into the financial and organizational relationships among different legal entities surrounding NRENs. Different funding approaches for covering NREN operating costs were analyzed (user fees versus government subsidies). Two major pricing schemas (flat rate and usage-based) were considered together with their advantages and disadvantages for NRENs.

The authors will continue the research work, initially presented in this paper, by carrying out the further investigation of the problem area (management of research and education networks). Any remarks and comments concerning the paper and the research project will be greatly appreciated. For the most current information about the research please refer to the project's homepage.

7. Acknowledgments

The authors would like to thank the reviewers of the paper: Dr. Yurij Demchenko (TERENA), Dr. Vladimir Galagan (National Technical University of Ukraine), and ir. Ilian Ilkov (Delft University of Technology).

The authors would like to express a special acknowledgment to Yurij Demchenko for his continuous support of the research.

8. References

1. Materials of NATO/CEENet/TERENA Advanced Networking Workshop, "Management and Policy Issues in running a National Research and Education Network," Yaroslavl, Russia, 13-16 June 1998, http://www.terena.nl/conf/nato-yar/, http://www.yars.free.net/conf/nato-yar/.
2. Maarten Looijen, Information Systems Management, Control and Maintenance, 1998, Kluwer BedrijfsInformatie b.v., Deventer
3. John Dyer, UK Academic Network Architecture: A Study commissioned by the ACN, http://www.ja.net/documents/net_arch.html
4. Memorandum of Understanding Concerning Levels of Service, UKERNA/JANET, http://www.ja.net/documents/sla.html
5. Proceedings of Internet Economics Workshop, March 1995, http://www.press.umich.edu/jep/econTOC.html
6. Florencio I. Utreras, REUNA: How an Academic Network can be Self Funded, http://www.isoc.org/inet95/proceedings/PAPER/121/html/paper.html
7. Nevil Brownlee, New Zealand Experiences with Network Traffic Charging, http://www.press.umich.edu:80/jep/works/BrownNewZe.html, http://www.auckland.ac.nz/net/Accounting/nze.html
8. Links to European Internet eXchanges, http://www.uni-c.dk/dix/euro.html, http://www.isi.edu/div7/naps/naps_eu.html
9. The Web-sites of European NRENs: ACONET (Austria), BELNET (Belgium), DARENET (Denmark), RENATER (France), DFN (Germany), GRNET (Greece), HUNGARNET (Hungary), HEANET (Ireland), GARR (Italy), SURFNET (Netherlands), NORDUNET, RCCN (Portugal), ARNES (Slovenia), REDIRIS (Spain), SUNET (Sweden), SWITCH (Switzerland), JANET (UK).
10. The Web-site of SINET (Japan) http://www.nacsis.ac.jp/

9. Author information

Maarten Looijen is a professor at the Information Strategy and Management of Information Systems Department at Delft University of Technology (Faculty of Information Technology and Systems). He has many years of experience in the management of computing centres and the development of information systems, and has published many books and articles on these subjects (see the author's homepage).

Dmitry Galagan is a postgraduate (Ph.D.) student in the same department (see the author's homepage). He is carrying out a research project with a provisional title "Modelling Management of Research, Education and Academic Networks." Maarten Looijen is the supervisor (promotor) of this research project.