Developing Networks in Less Industrialized Nations

Larry Press

IEEE Computer, vol 28, No 6, June, 1995, pp 66-71


Computer-mediated communication is becoming an important part of the global communication infrastructure. There are academic and research networks, commercial services, and proprietary networks. All are growing, and gateways between them are proliferating.

The Internet, the major research and education network, grew from the 4-node ARPANET in 1969 to around 17,979 networks in 60 nations, with 2,217,000 hosts as of January, 1994, according to Mark Lottor, who has programmed a "walk" of the Internet. Lottor found that the number of Internet hosts grew by 69% during 1993. While not as dramatic, commercial services are also growing rapidly. Market researchers at the International Data Corporation estimate the 1991 email market for North American carriers as $438.88 million, and they predict growth to $1,718 million by 1996. Proprietary networks are also exploding. For example, DEC's internal network grew from less than 10,000 hosts in 1986 to 83,410 today, and it connects 542 locations in 37 nations. Today most large organizations have wide-area networks, and carriers are developing and deploying new services which will facilitate the interconnection of local area networks by smaller companies.

The growth of the Net {footnote 1} is good news. The bad news is that it is concentrated in the Northern Hemisphere: North America, Western Europe and parts of Asia (see Figure 1). Less industrialized nations (LINs){footnote 2} are sparsely connected. Furthermore, connectivity is uneven within industrial nations. For example, children in some Latin American schools have access to the Net, while those in the inner cities of the US seldom do.

But there is more good news. Since communication infrastructure in LINs is typically poor, networks can have a significant positive impact, and continuing technological improvement means a meaningful network can be built using low-cost, appropriate technology such as personal computers and voice-grade, switched communication lines.

Relcom

For example, consider the evolution of the Relcom (RELiable COMmunications) network in the ex-Soviet Union. Relcom is a spin-off of the Demos Cooperative, which in 1982 began distributing a version of Unix in the Soviet Union. (Demos is a Russian acronym for "interactive portable operating system"). Demos initially built a small store-and-forward network using UUCP, the Unix to Unix Copy Program, in support of their users, but it became clear there was pent-up demand for communication outside their customer base, so Relcom was formed with the cooperation of the prestigious Kurchatov Institute of Atomic Research. Demos and Relcom were rather capitalistic from their inception -- without government sponsorship. Relcom became a commercial enterprise in April, 1989, and is currently independent of, though cooperative with, Demos.

On August 22, 1990, Relcom established a dial-up, store-and- forward link with EUnet through Helsinki Finland. This enabled them to exchange electronic mail with the rest of the world, and to send and receive Usenet News. An hourly dial-up connection was made to Helsinki for the batch transfer of mail and news. By the time of the Soviet coup attempt in August, 1991, Relcom had grown to 391 organizations in 70 cities ranging from Saint Petersberg in the west to Vladivostok in the east, and it played a significant role during the coup attempt. [7]

Since that time, Relcom has grown rapidly. As of September, 1993, Relcom had nearly 7,000 organizations and an estimated 200,000 users [11]. The network ties 162 regions and cities, and there are regional network centers in cities in Russia (57 cities), Byelorussia (3), Georgia (1), Kirghizstan (1), Latvia (1), Lithuania (3), Uzbekistan (1), and Ukraine (3).

At the time of the coup attempt, nearly all of the communication within Relcom was done with switched voice lines. Today leased lines connect the regional network centers. A persistent connection was established between Moscow and Rotterdam in October, 1992. That quadrupled the capacity for transfer between Relcom and the rest of the world and made interactive sessions possible. In 1993 channels from St. Petersberg and Talin were also opened.

While it has been successful, Relcom is not like a national network in an industrial nation. Where a network in the US typically uses large computers and workstations connected with dedicated high speed communication links, Relcom uses mostly personal computers and dial-up telephone connections. As of July 7, 1992, there were 2,601 nodes, 2,587 of which were PCs, with at least 2,444 running DOS. The other machines were microvaxes, unix workstations, and a lone Mac.{footnote 3}. They used a DOS- based version of UUCP on machines too small to run Unix. It was not possible to update these figures exactly, but the current network has added IP connectivity in some locations (accounting for less than 1% of the users), and larger PCs are becoming common [11].

Except for the few with IP connectivity, Relcom users are restricted to electronic mail and news. They cannot transfer files and log onto remote computers. This limitation is unfortunate, but not crucial. A study of 908 scientists in Chile, where IP connectivity is commonly available, found 1 in 4 were Net users, 1 in 8 joined listservers or made file transfers, and 1 in 20 used remote databases [9]. While all should aspire to "full" connectivity (a moving target), economic considerations will always dictate heterogeneous service. It is fortunate that email is both the lowest cost, and the most valuable service.

Relcom has grown rapidly using low-cost computers and modems and dial-up telephones. However, there is variance within and among nations. Relcom's growth was facilitated by factors like a relatively large installed base of personal computers, the availability of well trained technical people with a ten-year entrepreneurial history, relative freedom from regulation, and an adequate telephone system.

In other nations, the network may grow more slowly, but the bottom line is that there is a significant opportunity. A relatively small investment in people and equipment can improve communication within a nation or region, and once an internal network is established, it can be connected to the Net. Doing so will facilitate communications within LINs as well as between LINs and the industrial nations.

Resources and Applications

This section presents network-related resources and applications in LINs. During this discussion, we must remember that each case is different. For example, Relcom uses similar technology to and shares a lack of capital with the much smaller Cuban networks, but the communication infrastructures, user and technician pools, and applications are quite different. Furthermore, computer networks (in any nation) are shaped and constrained by communication, economic and education policies. We will consider communication infrastructure, software, hardware, people, and sponsorship and applications.

Communication Infrastructure

Good communication infrastructure makes networking easier, but also makes it less necessary. While networks in industrial nations use high-speed digital communication lines, networks in LINs use voice-grade phone lines. Switched telephone connections are used for communication between network nodes, between users and network nodes, and for communication with the world networks. For example, at the time of the coup attempt, Relcom had only two leased lines.

Communication is generally better in urban areas than in the countryside. Relcom finds calls within Moscow relatively reliable, but intercity calls within Russia can be difficult. (I spent many hours trying to make a reliable UUCP connection between St. Petersberg and Moscow during the summer of 1992). The same holds true for much of Africa and Latin America. Phone density also varies greatly between nations. For example, it is estimated that the number of phones per capita varies by a factor of two in Eastern Europe (Table 1).

While generally high, phone charges vary. For example, when Relcom began, calls were cheap in the Soviet Union. Moscow phone service had a low, fixed rate, making all-day dial-up connections feasible, and calls from Relcom headquarters to Helsinki were only 6 rubles per minute. The market price of phone connections in the ex-Soviet Union is rising rapidly.

International calls to and from LINs are often very expensive. For example, the Cuban networks pay approximately $700 per month for communication with Canada, their entry point to the Net. At that rate, the 386-based PC that makes the connection would cost around two months phone bill. (In spite of the embargo, popular U. S. and Asian personal computers are readily available in Cuba for around 1.5 times U. S. prices). A call from Zambia to the U. K. costs $7/minute (a professor's salary is approximately $100/month), and the successful completion rates are as low as 12% on international calls from Ghana. Calls from Latin America to the U. S. cost from $2-6/minute. As such, international calls to link to the world networks are often made from developed nations where they are cheaper and easier to place.

In some cases, packet radio communication may be necessary. Volunteers in Technical Assistance (VITA), has begun packet radio experiments using "microsatellites," [4]. Satellife, a health organization, has used the same technology for medical information retrieval, mail and conferencing. They hope to connect 100 medical libraries, ministries of health, non- governmental agencies, and research institutes. Motorola and several other companies are deploying worldwide satellite-based networks which will permit voice and data communication in remote and undeveloped regions; however, these links may be expensive.

The success and spread of computer networks is linked to the cost, quality and availability of the communication infrastructure. These impacts should be considered when formulating policies on communication regulation and investment [5]. Communication infrastructure also influences computer proliferation, which is discussed in the following section.

Hardware

PCs running DOS or UNIX are the most common hardware in LIN networks, though one sees occasional UNIX workstations. The cost of suitable computers is low and continues to fall.

Since capital is scarce in LINs, it may be tempting to use minimal hardware, but this is probably false economy. Relatively powerful hardware makes it possible to use more powerful, standardized software, saving time and training costs. Furthermore, there is a tradeoff between hardware and communication costs. Phone connections are often unreliable, noisy and expensive. In such situations, a variable-rate, error- correcting, modem that compresses data is a bargain, saving communication cost and operator time.

While capital is limited everywhere, there is significant variance in the installed base of PCs. Part of the rapid growth of Relcom is attributable to the fact that there were relatively many unconnected computers in the country (see Table 2). Other nations, for example in Africa, have fewer PCs and lower PC density.

In spite of technological progress, nations often adopt policies which inhibit the proliferation of computer technology. Tariffs, regulations, restrictions on currency exchange, and other measures intended to protect local industry and raise revenue, have the undesirable side effect of raising the cost of computers and communications. For example, in surveying the Indian software industry, Yourdon [12] tells us "the history of the Indian computer industry can be neatly divided into two phases: a stagnant, IBM-less, pre-PC period, and a new, thriving era that began with the introduction of the PC and [Rajiv] Gandhi's push for modernization of the information technology sector." Latin American nations such as Brazil and Chile have also liberalized computer (and other) trade after protectionist eras [2]. Once again, we see that computer networks are part of a larger socio- technical system.

Software

Two types of software are needed in store-and-forward networks, one for the exchange of batches of information between network nodes and another for interaction between a user and a network node. While most of the commonly used software is in the public domain, it is often obsolete and needlessly difficult to use.

The software used between nodes is usually UUCP or Fido. Both are available for zero or little cost. They operate in store- and-forward mode, collecting mail and other messages in batches until a connection is established to another node. At that time, incoming and outgoing messages are exchanged. The batches of data are compressed to reduce connect time. In addition to mail, most LIN networks participate in the world-wide Usenet news system. Again, the software costs little or nothing.

End users also require software to read, send, and file mail and News. Again, public domain and low cost software is available. For example, Kermit, a public domain terminal emulation program, is available for many platforms, and has been distributed systematically in Eastern Europe for several years. There was even an international conference on Kermit in Moscow.

While current software is available at low cost, there is much room for improvement. Installing and maintaining networking software like UUCP or Fido requires considerable expertise. End- user software is also complex and cumbersome. For example, to send or receive mail, one must learn to use a terminal emulation program like Kermit and a mail program and editor on the network host. You also need some familiarity with the operating systems of the host and terminal computers, and a conceptual model of the flow of information within and between the two systems. The problem is further exacerbated by the fact that most software and documentation is written in English. (English is also the predominant language in most Usenet News groups and in mailing lists).

End user software is confusing and system software is difficult to install and tune. Software designed to run on a computer (as opposed to emulating a terminal) with a modern user interface would reduce training costs and help spread connectivity [8]. Commercial packages such as Mortice Kern Systems' "Internet Anywhere," may fill this gap if priced reasonably in LINs.

People

As with software, there are two classes of people, system administrators and users. Systems administrators are computer professionals. As we saw, Relcom was established by members of the Demos Cooperative, which had been selling and supporting a version of Unix since 1982. The Relcom professional staff are highly trained graduates of excellent Russian universities, but they are not enough. Each regional network center requires a system administrator familiar with Unix, and this has been a key constraint on their growth according to co-founder Valerij Bardin. The situation is more critical in most LINs.

One solution to the personnel problem has been to rely on visiting experts from developed nations. While commendable, such efforts are primarily of value in the early, technical stages, and they must eventually be phased out in favor of local self sufficiency.

When a network is first being established, technical skills are paramount, but once it begins to grow, managerial and financial expertise are also needed. Relcom began as a cooperative, but they have now converted to a limited organization, and it will be interesting to see if they can attract capital and develop management and marketing skills for the transition from a start- up to a large, stable organization.

Training users is more difficult than training system administrators and managers. While training an individual user is easy, widespread networking requires a large base of sophisticated, demanding users. Cuba has taken a noteworthy approach to this problem [6]. They have established a nationwide organization of Youth Computer Clubs (YCCs). In February, 1992 there were YCCs in over 130 cities. They are reminiscent of Bob Albrecht's People's Computer Company (PCC) and similar experiments dating back to the 1960s in the United States. Like the PCC, they have computers running games, drawing programs, and other software, which the children may use in a relatively unstructured manner. Additionally, the YCCs offer classes on using professional application packages and programming. Advanced classes cover sophisticated topics such as C++.

The Cuban YCCs also operate a network using a 386-based PC with a 2400-baud modem running Unix in their headquarters in Havana. Over 30 of their locations now have computers with modems, which are used to dial Havana to send and receive mail. The YCC computer in Havana transfers their mail to another PC in Havana which exchanges mail daily with Canada. Thus children and adults in rural Cuba can exchange email with people throughout the world.

A nation's educational policy and goals are closely tied to the success of its computer networks. Support for and quality of university computer science programs varies greatly among LINs. Some, for example Chile [1], have ambitious programs and hope to establish world-class research institutes, while others have no computer science education.

There are also great discrepancies in pre-university education. Ivan Illich and others have pointed out that education is often used to maintain the social status quo. Nations such as Cuba, Malaysia, and Costa Rica have ambitious computer literacy programs, which hope to reach all people -- rich, poor, rural, and urban. Differences between public and private education and urban and rural education exist in all nations, and those differences will in part determine who, and how many people use computer networks.

Sponsorship and Applications

Sadowsky [10] states that "individuals and organizations working toward international connectivity for their countries need to demonstrate how having such connectivity will address directly the needs of government ministries and will contribute to the development objectives of the country." Cuba provides an example of government sponsored networking [6].

Relcom provides a contrast. In spite of having begun under a communist regime, they did not seek government support (though government agencies are their customers). Table 3 shows the variety of Relcom users (and hence sponsors) as of August, 1991. There were users in government agencies, foreign and domestic publications and news services, commodity, raw material and stock exchanges, universities and university departments, limiteds, corporations, enterprises, companies, firms or banks, joint ventures or small ventures, and scientific and research institutes. It was not possible to update Table 3, but the same general variety exists today [11].

Having a broad range of users and applications has been a source of stability and independence for Relcom, where many users are commercial enterprises. Relcom co-founder Vadim Antonov feels that the social and business impact of Relcom has been greater than that of the Internet in the US because they were open to commercial service from their inception. He points out that Yeltsin had electronic mail 2 years before the US President, and says the network has become Russia's largest marketplace. Networks in LINs typically serve universities and government and non-government organizations, but network developers would do well to involve business users. Commercial connectivity will be even more important as the Internet becomes more widely commercialized.

Summary

Low cost, appropriate technology networks can make significant additions to the communication infrastructure of LINs, and link them to the rest of the world. We have described the growth of Relcom, and surveyed some of the resources which are needed for such networks. In doing so, we have noted that there is considerable variability among nations, and that government policies in areas such as education and trade have significant impact on computer networks. Readers wishing further reading or discussion on this topic are referred to an archive of papers and other documents, available via anonymous ftp from dhvx20.csudh.edu (155.135.1.1). The directory is called "global_net". Readers in LINs who are interested in learning more about network connectivity, can contact the Internet Society, which conducts an annual hands-on workshop on the topic.


References

1. Baeza-Yates, Ricardo A., Fuller, David A., Pino, Jose, Innovation as a Critical Success Factor for the Development of an Information Technology Industry in Chile, Proceedings of IFIP '92, Madrid.

2. Beli, Pedro, Globalizing the Rest of the World, Harvard Business Review, July-August, 1991, pp 50-55.

3. Dyson, Esther, Eastern Europe Trip Report, Release 1.0, May 31, 1990, EDventure Holdings, New York, pp 1-30.

4. Garriott, Gary L., Packet Radio in Earth and Space Environments for Relief and Development, Proceedings of the 34th Annual Meeting of the African Studies Association, Saint Louis, MO, November 23-26, 1991.

5. Goodman, S, Press, L, Ruth, S., and Rutkowski, A., "The Global Diffusion of the Internet: Patterns and Problems," Communications of the ACM, in press.

6. Mesher, G., Briggs, R., Goodman, S, Press, L., and Snyder, J., "Cuba, Communism, and Computing," Communications of the ACM, November, 1992.

7. Press, L., "Relcom, An Appropriate Technology Network," Proceedings of INET '92, International Networking Conference, Kobe, Japan, June, 1992, Internet Society, Reston, VA. Reprinted in "The Proceedings of the Telecommunications Conference," Moscow, Russia, June, 1992.

8. Press, L., "Empowering Low-Bandwidth Users," Proceedings of INET '93, International Networking Conference, San Francisco, August, 1993, Internet Society, Reston, VA.

9. Ruth, S., Goet, R., "Must Invisible Colleges be Invisible? An Approach to Examining Large Communities of Network Users," Internet Research, 3, 1 (Spring 1993), 47.

10. Sadowsky, G. "The Importance to Governments of Access to Distributed Knowledge," Proceedings of INET '92, International Networking Conference, Kobe, Japan, June, 1992, Internet Society, Reston, VA.

11. Shliemin, Vladimir, personal communication, September, 1993.

12. Yourdon, Ed, India, American Programmer, October, 1989, pp 3- 26, New York.

Footnotes

1. We will use the term "the Net" to refer to the Internet and all networks with at least email connectivity to the Internet.

2. We use "LIN" instead of the more common "developing nation," because, as measured by the United Nations Development Program human development is a process of enlarging people's choices. Given a comprehensive Human Development Index, UNDP concludes that world development gaps have narrowed as income gaps have widened.

3. The statistics in this paragraph were derived from an analysis of Relcom's UUCP map, the list of nodes which governs message routing, in July, 1992.


Tables

          Country        Population    Phones     Phones
                         (millions)     (000) (/100 people)
          -------------------------------------------------
          Bulgaria             9.0      1,800       20
          Czechoslovakia      15.6      3,588       23
          E. Germany          16.6      3,486       21
          Hungary             10.6      1,378       13
          Poland              40.0      4,800       12
          Rumania             23.1      3,003       13
          USSR               287.6     34,512       12
          Yugoslavia          24.0      2,880       12
          United States      250.0    162,500       65
Table 1. Eastern European Phones [3]. While the figures are rough estimates, they illustrate regional heterogeneity.
-----

          Country        Population   DOS PCs     People
                         (millions)     (000)     per PC
          ----------------------------------------------
          Bulgaria             9.0        100         90
          Czechoslovakia      15.6         60        260
          E. Germany          16.6         50        332
          Hungary             10.6         60        177
          Poland              40.0        100        400
          Rumania             23.1          2      11550
          USSR               287.6        350        822
          Yugoslavia          24.0         75        320
          United States      250.0      35000          7
Table 2. Eastern European PCs [3]. While the figures are rough estimates, they illustrate regional heterogeneity.
-----

11  government agencies, including the USSR and Russian Finance 
    Ministries

15  foreign and domestic publications and news services including 
    AP, UPI, the German Press Agency, and Financial Times

20  commodity, raw material and stock exchanges  

26  universities and university departments, including several 
    machines at Moscow State and Saint Petersberg Universities

96  limiteds, corporations, enterprises, companies, firms or 
    banks, joint ventures or small ventures

117 scientific and research institutes, nearly all in technical 
    fields such as mathematics and physics.
Table 3. Relcom Organizations. The count for each category was based on organization names. I was unable to classify 85 organizations, but the degree of diversity is illustrated.

Figure Caption

Figure 1. Global Connectivity. This map shows that both the Internet and store-and-forward networks are distributed unevenly around the world. The map is one of a series of regular publications of John Quarterman of Matrix, Inc., jsq@tic.com.

Abstract

The global computer-mediated communication network is growing rapidly, but it is concentrated in the affluent parts of industrialized nations. While this could widen the income gap between industrialized and less industrialized areas, it also presents an opportunity since low cost, appropriate technology networks can make significant additions to the communication infrastructure of less industrialized areas, and link them to the rest of the world. We describe the evolution and growth of one such network, Relcom in the ex-Soviet Union. This is followed by a discussion of pertinent resources: communication infrastructure, hardware, software, and people. Applications and sponsorship are also covered. Throughout, we note differences among nations and the effect of government policy on networks.

Keywords

Internet, WAN, Network, Global Development, Communication