Stephen Ruth <ruth@gmu.edu>
George Mason University
USA
Randy Bush <randy@nsrc.org>
University of Oregon
USA
This paper gives a brief sketch of a practical approach to network planning in a large developing nation--Indonesia. Even though the approach is unremarkable and taught in most management schools, in practice it frequently does not succeed. The authors describe a bottom-up sequence that begins with determining end-user needs and gradually works toward creating a system that meets those needs. As the process has been successful (implementation planning is underway) there are several insights that may guide others faced with the same challenge. Among them are careful collaboration with local and regional specialists, focusing on the needs of the end user, and requiring that equipment and structural decisions be subordinate to management strategy.
In the fall of 1995 the Government of Indonesia (GOI) was in the process of a major overhaul of its approach to developing a scientific network. We were given the opportunity to be a part of GOI's effort and include below some of the essential characteristics of the approach. A more detailed description is to be found in our article in a forthcoming issue of the journal Information Technology for Development. The goal was to catalyze the current Indonesian Scientific Network, called IPTEKnet, from a cluster of disaggregated networks to be a full-service provider for the entire science community. As consultants to GOI and the World Bank, we approached the problem in exactly the way we teach it in class. First, we recommended that the approach be one that subordinated structure to strategy (Ruth, 1996). By gaining acceptance of that idea we excluded any considerations of equipment, contractors, certifications, buildings, salaries, and so on until we were able to learn exactly what the government's strategy was and how the end users were being considered. After clarifying those issues, it was possible to proceed quickly.
The government's strategy was to examine several models of network diffusion and to select the one that was most in keeping with national policy goals (Press, 1996). At one end of the spectrum of options was a completely government-funded system, somewhat similar to NSFNET, to act as a catalyst for encouraging public and private-sector development (Goodman, et al., 1995). The other extreme would be a complete privatization of the scientific networking function. An intermediate option, the subject of this description, is a hybrid organization that both serves the scientific users and gains revenues from private-sector users.
Once we understood the government's willingness to countenance this hybrid organization--one capable of offering services to government scientists and also of creating a revenue stream to offset costs--it was possible to have a clear model of the structure that would be necessary.
The approach was straightforward--almost a case study in break-even analysis with bottom lines, sensitivity analysis of critical variables, and a wide variety of scenarios to be considered. First we determined the location of the users--spread throughout the mainland plus Irian Jaya, Borneo, and more than 13,000 mostly very small islands. The initial estimate was that about 15,000 persons were the scientific network's main user base, at least as a beginning. Table 1 shows the distribution of these users based on the large regions in Indonesia. The largest of these regions, Jakarta, has approximately 1,300 users for the scientific network. These estimates were based on the disposition of scientific activity and personnel in the Jakarta region. Similar data were available for 11 other regions aggregating 13,200 scientific users for the network. It was stipulated that most of these users would eventually require enough bandwidth to pull down WWW for at least an hour per day and have good e-mail connectivity at all times. The columns show gradual accretion of new users over six-month intervals from October 1995 until October 1999. The table also indicates the allocations of points of presence (POP) throughout the 11 regions. Three levels of equipment were selected on the basis of projections of user needs. For a large POP, two routers, a communications server, DSU, modems, and an uninterruptible power supply were needed along with other miscellaneous equipment and supplies--approximate cost, $40,000. A medium POP requires a router, server, DSU, modems, and other equipment--approximate cost, $30,000. For a small POP only a communications server, DSU, and modems were necessary--approximate cost, $6,000. We allocated totals of 14 large POPs, 44 medium POPs, and 27 small POPs as the major fixed equipment investment.
Zone | Oct-95 | Apr-96 | Oct-96 | Apr-97 | Oct-97 | Apr-98 | Oct-98 | Apr-99 | ||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
B | M | S | L | M | S | L | M | S | L | M | S | L | M | S | L | M | S | L | M | S | L | M | S | Total | ||
Jakarta | 0 | 1 | 2 | 1 | 4 | 1 | 4 | 1 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 25 | |||||||
Bandung | II | 1 | 1 | 1 | 2 | 1 | 2 | 1 | 1 | 10 | ||||||||||||||||
Jogja | IV | 1 | 2 | 1 | 1 | 1 | 6 | |||||||||||||||||||
Surabaya | IV | 1 | 2 | 1 | 1 | 1 | 6 | |||||||||||||||||||
Batam | IV | 1 | 2 | 1 | 2 | 1 | 2 | 1 | 2 | 1 | 13 | |||||||||||||||
Denpasar | III (via Sby) | 1 | 1 | 1 | 1 | 4 | ||||||||||||||||||||
U.Pandang | IV | 1 | 2 | 1 | 1 | 1 | 6 | |||||||||||||||||||
Medan | V | 1 | 1 | 1 | 3 | |||||||||||||||||||||
Menado | IV (via UP) | 1 | 1 | 1 | 3 | |||||||||||||||||||||
Semarang | II (via Jgj) | 1 | 1 | 1 | 3 | |||||||||||||||||||||
Palembang | IV | 1 | 1 | 1 | 3 | |||||||||||||||||||||
B. Papan | IV | 1 | 1 | 1 | 3 | |||||||||||||||||||||
2 | 3 | 0 | 5 | 6 | 2 | 2 | 12 | 5 | 3 | 8 | 6 | 2 | 4 | 4 | 0 | 5 | 5 | 0 | 3 | 3 | 0 | 3 | 2 | 85 | ||
Total | Total | 5 | 18 | 37 | 54 | 64 | 74 | 80 | 85 | |||||||||||||||||
Large | 2 | 7 | 9 | 12 | 14 | 14 | 14 | 14 | ||||||||||||||||||
Medium | 3 | 9 | 21 | 29 | 33 | 38 | 41 | 44 | ||||||||||||||||||
Small | 0 | 2 | 7 | 13 | 17 | 22 | 25 | 27 |
Tables 2 and 3 describe the location and size of the dial-up and leased-line users. The typical user will utilize the system for less than 15 hours per month, according to current IPTEKnet records and the experiences of local providers in Jakarta. Assumptions about user behavior were based on discussions with many local providers in Jakarta, analysis of published rates, and review of experience with current IPTEKnet subscribers, individual and institutional. Data were also gathered for leased-line segments across the country's regional nodes with bandwidth at 19.2 or 64 kilobits per second (kbps). International costs of satellite connection from Jakarta, which quickly moves from 128 kbps (about $100,000 per six months) to 256 kbps ($180,000) to 512 kbps ($250,000), are partly subsidized by GOI.
Oct-95 | Apr-96 | Oct-96 | Apr-97 | Oct-97 | Apr-98 | Oct-98 | Apr-99 | Total | ||
---|---|---|---|---|---|---|---|---|---|---|
Jakarta | Jakarta | 200 | 150 | 150 | 100 | 100 | 700 | |||
Bogor | 30 | 30 | 40 | 100 | 200 | |||||
Tanggerang | 30 | 30 | 40 | 100 | 200 | |||||
Bekasi | 30 | 30 | 40 | 100 | 200 | |||||
Bandung | Bandung | 200 | 150 | 150 | 100 | 100 | 700 | |||
Tasikmalaya | 30 | 30 | 40 | 100 | 200 | |||||
Cirebon | 30 | 30 | 40 | 100 | 200 | |||||
Sukabumi | 30 | 30 | 40 | 100 | 200 | |||||
Jogja | Jogja | 200 | 150 | 150 | 100 | 100 | 700 | |||
Magelang | 30 | 30 | 40 | 100 | 200 | |||||
Kudus | 30 | 30 | 40 | 100 | 200 | |||||
Solo | 30 | 30 | 40 | 100 | 200 | |||||
Surabaya | Surabaya | 200 | 150 | 150 | 100 | 100 | 700 | |||
Madiun | 30 | 30 | 40 | 100 | 200 | |||||
Cepu | 30 | 30 | 40 | 100 | 200 | |||||
Malang | 30 | 30 | 40 | 100 | 200 | |||||
Batam | Batam | 200 | 150 | 150 | 100 | 600 | ||||
Dumai | 30 | 30 | 40 | 100 | 200 | |||||
T. Pinang | 30 | 30 | 40 | 100 | 200 | |||||
Pekanbaru | 30 | 30 | 40 | 100 | 200 | |||||
Denpasar | Denpasar | 200 | 150 | 150 | 100 | 600 | ||||
Mataram | 30 | 30 | 40 | 100 | 200 | |||||
U. Pandang | U. Pandang | 200 | 150 | 150 | 100 | 100 | 700 | |||
Pare-pare | 30 | 30 | 40 | 100 | 200 | |||||
Goa | 30 | 30 | 40 | 100 | 200 | |||||
Tanah Toraja | 30 | 30 | 40 | 100 | 200 | |||||
Ambon | 30 | 30 | 40 | 100 | ||||||
Jayapura | 30 | 30 | 40 | 100 | ||||||
Medan | Medan | 200 | 150 | 150 | 100 | 100 | 700 | |||
Pematang S. | 30 | 30 | 40 | 100 | 200 | |||||
Benda Aceh | 30 | 30 | 40 | 100 | 200 | |||||
Manado | Manado | 200 | 150 | 150 | 100 | 100 | 700 | |||
Ternate | 30 | 30 | 40 | 100 | 200 | |||||
Semarang | Semarang | 200 | 150 | 150 | 500 | |||||
Salatiga | 30 | 30 | 40 | 100 | ||||||
Palembang | Palembang | 200 | 150 | 150 | 100 | 600 | ||||
Pangkal P. | 30 | 30 | 40 | 100 | 200 | |||||
T. Karang | 30 | 30 | 40 | 100 | 200 | |||||
Padang | 30 | 30 | 40 | 100 | 200 | |||||
B. Papan | B. Papan | 200 | 150 | 150 | 500 | |||||
Samarinda | 30 | 30 | 40 | 100 | ||||||
Pontianak | 30 | 30 | 40 | 100 | ||||||
0 | 580 | 1350 | 1750 | 2810 | 3030 | 1980 | 1700 | 13200 | ||
580 | 1930 | 3680 | 6490 | 9520 | 11500 | 13200 |
Oct-95 | Apr-96 | Oct-96 | Apr-97 | Oct-97 | Apr-98 | Oct-98 | Apr-99 | |
---|---|---|---|---|---|---|---|---|
Large | 2 | 7 | 9 | 12 | 14 | 14 | 14 | 14 |
Medium | 3 | 9 | 21 | 29 | 33 | 38 | 41 | 44 |
Small | 0 | 2 | 7 | 13 | 17 | 22 | 25 | 27 |
Total | 5 | 18 | 37 | 54 | 64 | 74 | 80 | 85 |
Total users | 1000 | 3370 | 5320 | 7230 | 8370 | 8920 | 9250 | 9570 |
Note: Estimates based on 350 users for a large institution, 100 for a medium-size institution, and 10 for a small institution.
The collection of both cost and revenue information in the form of a spreadsheet is relatively simple. Once the user locations, POP costs, and dial-up and leased-line costs are collected and projected over the eight half-year intervals, it is easy to represent a wide range of cases for review (World Bank, 1996). Total costs consist of equipment, telcom, and administrative expenses. Total revenues are the sum of payments from government for scientific users and from private users for the dial-up service. Table 4, for example, assumes that fixed equipment costs are amortized in a relatively short time--three years--and that the number of leased-line and dial-up users is close to the projections: 13,200 and 9,570, respectively.
The bottom line (literally) of this spreadsheet makes it possible to determine the point in time when the project begins to pay back all its costs and operate as a financially viable concern, having returned all its expenses from earlier years. In this case that point is achieved at the end of three years, as the difference between aggregate revenues and costs turns positive. A less stringent, but far more commonly used, calculation is the point where current operations operate in the black; that is, when the period revenues begin to exceed the period costs. For the standard case this point is reached at the end of the second year (when the next to last row from the bottom turns positive).
Even under these relatively pessimistic assumptions about potential revenues, equipment costs, and regional and international line charges, the new IPTEKnet begins to be self-sustaining in about two years and pays back all costs in three years. This occurs even though the equipment costs are expensed immediately, rather than being amortized over a long period.
Oct-95 | Apr-96 | Oct-96 | Apr-97 | Oct-97 | Apr-98 | Oct-98 | Apr-99 | ||
---|---|---|---|---|---|---|---|---|---|
Leased line | 35,114 | 182,182 | 232,227 | 348,273 | 420,818 | 420,818 | 420,818 | 420,818 | |
Modem cost | 0 | 4,364 | 9,818 | 12,727 | 20,364 | 22,182 | 14,545 | 12,364 | |
Phone line installation | -- | 5,455 | 12,273 | 15,909 | 25,455 | 27,727 | 18,182 | 15,455 | |
Monthly phone | 764 | 1,909 | 4,486 | 7,827 | 13,173 | 18,995 | 22,814 | 26,059 | |
International lines | 103,500 | 177,205 | 177,205 | 252,477 | 252,477 | 252,477 | 252,477 | 252,477 | |
Salary | 65,455 | 70,909 | 81,818 | 92,727 | 109,091 | 114,545 | 120,000 | 125,455 | |
Travel | 1,364 | 4,545 | 5,909 | 5,909 | 10,455 | 10,455 | 10,455 | 10,455 | |
Fixed equipment | 134,427 | 156,114 | 48,827 | 133,955 | 91,391 | -- | -- | -- | |
Recurring equipment | 1,364 | 4,909 | 10,091 | 14,727 | 17,455 | 20,182 | 21,818 | 23,182 | |
Miscellaneous | 73,182 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
Total cost | 415,168 | 607,591 | 582,655 | 884,532 | 960,677 | 887,382 | 881,109 | 886,264 | |
Dial-up | Incremental fee | -- | 791 | 1,841 | 2,386 | 3,832 | 4,132 | 2,700 | 2,318 |
Starting fee | -- | 13,182 | 30,682 | 39,773 | 63,864 | 68,864 | 45,000 | 38,636 | |
Monthly | -- | 63,273 | 210,545 | 401,455 | 708,000 | 1,038,545 | 1,254,545 | 1,440,000 | |
Leased line | |||||||||
Starting fee | Large | 1,818 | 4,545 | 1,818 | 2,727 | 1,818 | -- | -- | -- |
Medium | 1,364 | 2,727 | 5,455 | 3,636 | 1,818 | 2,273 | 1,364 | 1,364 | |
Small | -- | 455 | 1,136 | 1,364 | 909 | 1,136 | 682 | 455 | |
Monthly | Large | 10,909 | 38,182 | 49,091 | 65,455 | 76,364 | 76,364 | 76,364 | 76,364 |
Medium | 12,273 | 36,818 | 85,909 | 118,636 | 135,000 | 155,455 | 167,727 | 180,000 | |
Small | -- | 5,455 | 19,091 | 35,455 | 46,364 | 60,000 | 68,182 | 73,636 | |
Total income | 26,364 | 165,427 | 405,568 | 670,886 | 1,037,968 | 1,406,768 | 1,616,564 | 1,812,773 | |
Income - cost | (388,805) | (442,164) | (177,086) | (213,645) | 77,291 | 519,386 | 735,455 | 926,509 | |
(388,805) | (830,968) | (1,008,055) | (1,221,700) | (1,144,409) | (625,023) | 110,432 | 1,036,941 |
For this case the scenario of Table 4 is modified to zero out equipment costs. (See Table 5.) If the equipment is paid for by a grant or loan from a donor, there is no expensing required and the payback period is sometimes moot. Interestingly, zeroing out the equipment costs does not have a significant effect on the time required to repay the investment. It is a few months less than the standard case but not dramatically altered. This is because equipment costs are overpowered by line costs and are not a major determinant in the outcome. The network is self-sustaining after two years and pays for all costs in three years.
Oct-95 | Apr-96 | Oct-96 | Apr-97 | Oct-97 | Apr-98 | Oct-98 | Apr-99 | ||
---|---|---|---|---|---|---|---|---|---|
Leased line | 35,114 | 182,182 | 232,227 | 348,273 | 420,818 | 420,818 | 420,818 | 420,818 | |
Modem cost | 0 | 4,364 | 9,818 | 12,727 | 20,364 | 22,182 | 14,545 | 12,364 | |
Phone line installation | -- | 5,455 | 12,273 | 15,909 | 25,455 | 27,727 | 18,182 | 15,455 | |
Monthly phone | 764 | 1,909 | 4,486 | 7,827 | 13,173 | 18,995 | 22,814 | 26,059 | |
International lines | 103,500 | 177,205 | 177,205 | 252,477 | 252,477 | 252,477 | 252,477 | 252,477 | |
Salary | 65,455 | 70,909 | 81,818 | 92,727 | 109,091 | 114,545 | 120,000 | 125,455 | |
Travel | 1,364 | 4,545 | 5,909 | 5,909 | 10,455 | 10,455 | 10,455 | 10,455 | |
Fixed equipment | -- | -- | -- | -- | -- | -- | -- | -- | |
Recurring equipment | 1,364 | 4,909 | 10,091 | 14,727 | 17,455 | 20,182 | 21,818 | 23,182 | |
Miscellaneous | 73,182 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
Total cost | 280,741 | 451,477 | 533,827 | 750,577 | 869,286 | 887,382 | 881,109 | 886,264 | |
Dial-up | Incremental fee | -- | 791 | 1,841 | 2,386 | 3,832 | 4,132 | 2,700 | 2,318 |
Starting fee | -- | 13,182 | 30,682 | 39,773 | 63,864 | 68,864 | 45,000 | 38,636 | |
Monthly | -- | 63,273 | 210,545 | 401,455 | 708,000 | 1,038,545 | 1,254,545 | 1,440,000 | |
Leased line | |||||||||
Starting fee | Large | 1,818 | 4,545 | 1,818 | 2,727 | 1,818 | -- | -- | -- |
Medium | 1,364 | 2,727 | 5,455 | 3,636 | 1,818 | 2,273 | 1,364 | 1,364 | |
Small | -- | 455 | 1,136 | 1,364 | 909 | 1,136 | 682 | 455 | |
Monthly | Large | 10,909 | 38,182 | 49,091 | 65,455 | 76,364 | 76,364 | 76,364 | 76,364 |
Medium | 12,273 | 36,818 | 85,909 | 118,636 | 135,000 | 155,455 | 167,727 | 180,000 | |
Small | -- | 5,455 | 19,091 | 35,455 | 46,364 | 60,000 | 68,182 | 73,636 | |
Total income | 26,364 | 165,427 | 405,568 | 670,886 | 1,037,968 | 1,406,768 | 1,616,564 | 1,812,773 | |
Income - cost | (254,377) | (286,050) | (128,259) | (79,691) | 168,682 | 519,386 | 735,455 | 926,509 | |
(254,377) | (540,427) | (668,686) | (748,377) | (579,695) | (60,309) | 675,145 | 1,601,655 |
For another case the equipment cost is zeroed as before and to this is added the condition that the number of dial-up users added across the four years is increased by 20%. See Table 6. Because the potential market for dial-up users was already in the range of 100,000 in July 1995 and a few providers share a current pool of about 10,000 at most in the country, an increase in a four year projection from 13,000 dial-up users to 16,000 is not a major leap of faith. The network becomes self-sustaining in 1.5 years and repays all costs in less than 3 years.
Oct-95 | Apr-96 | Oct-96 | Apr-97 | Oct-97 | Apr-98 | Oct-98 | Apr-99 | ||
---|---|---|---|---|---|---|---|---|---|
Leased line | 35,114 | 182,182 | 232,227 | 348,273 | 420,818 | 420,818 | 420,818 | 420,818 | |
Modem cost | 0 | 4,364 | 9,818 | 12,727 | 20,364 | 22,182 | 14,545 | 12,364 | |
Phone line installation | -- | 5,455 | 12,273 | 15,909 | 25,455 | 27,727 | 18,182 | 15,455 | |
Monthly phone | 764 | 1,909 | 4,486 | 7,827 | 13,173 | 18,995 | 22,814 | 26,059 | |
International lines | 103,500 | 177,205 | 177,205 | 252,477 | 252,477 | 252,477 | 252,477 | 252,477 | |
Salary | 65,455 | 70,909 | 81,818 | 92,727 | 109,091 | 114,545 | 120,000 | 125,455 | |
Travel | 1,364 | 4,545 | 5,909 | 5,909 | 10,455 | 10,455 | 10,455 | 10,455 | |
Fixed equipment | -- | -- | -- | -- | -- | -- | -- | -- | |
Recurring equipment | 1,364 | 4,909 | 10,091 | 14,727 | 17,455 | 20,182 | 21,818 | 23,182 | |
Miscellaneous | 73,182 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
Total cost | 280,741 | 451,477 | 533,827 | 750,577 | 869,286 | 887,382 | 881,109 | 886,264 | |
Dial-up | Incremental fee | -- | 949 | 2,209 | 2,864 | 4,598 | 4,958 | 3,240 | 2,782 |
Starting fee | -- | 15,818 | 36,818 | 47,727 | 76,636 | 82,636 | 54,000 | 46,364 | |
Monthly | -- | 75,927 | 252,655 | 481,745 | 849,600 | 1,246,255 | 1,505,455 | 1,728,000 | |
Leased line | |||||||||
Starting fee | Large | 1,818 | 4,545 | 1,818 | 2,727 | 1,818 | -- | -- | -- |
Medium | 1,364 | 2,727 | 5,455 | 3,636 | 1,818 | 2,273 | 1,364 | 1,364 | |
Small | -- | 455 | 1,136 | 1,364 | 909 | 1,136 | 682 | 455 | |
Monthly | Large | 10,909 | 38,182 | 49,091 | 65,455 | 76,364 | 76,364 | 76,364 | 76,364 |
Medium | 12,273 | 36,818 | 85,909 | 118,636 | 135,000 | 155,455 | 167,727 | 180,000 | |
Small | -- | 5,455 | 19,091 | 35,455 | 46,364 | 60,000 | 68,182 | 73,636 | |
Total income | 26,364 | 180,876 | 454,182 | 759,609 | 1,193,107 | 1,629,076 | 1,877,013 | 2,108,964 | |
Income - cost | (254,377) | (270,601) | (79,645) | 9,032 | 323,821 | 741,695 | 995,904 | 1,222,700 | |
(254,377) | (524,978) | (604,624) | (595,592) | (271,771) | 469,924 | 1,465,827 | 2,688,527 |
This brief description gives the essentials of the approach. Because there were many options, not just one solution, the World Bank and GOI were able to integrate national policy issues with the various scenarios. The essential elements of this plan have been adopted. GOI has decided to integrate universities into the network in a much larger way than we had anticipated, centering much of the activity at five large institutions. Even these major increases in the scope and direction of the GOI plan are easily modeled in these spreadsheets.
We suggest three insights from this experience. First, by establishing a strategy before deriving the equipment structure, the client is able to have a policy-driven planning process. Once we understood the government's goals, the break-even analysis was straightforward. Second, both optimistic and pessimistic options must be clearly elaborated. In this case zeroing or not zeroing equipment costs proved to be relatively unimportant. The "drivers" were line costs and number of dial-up users. By showing the optimistic and pessimistic side of key variables we were able to offer GOI and World Bank decision makers a decision space that made their planning process simpler. Finally, studies of this type can sometimes go on for months, even years. We completed most of our work in three weeks. It will be seen that the data we gathered are not difficult to obtain and the process we used was not remarkable. We feel that the difference was the degree of cooperation from the Indonesian government and PTTs. But even without that cooperation, a study of this type can be done in a relatively short time using mostly unobtrusive sources of information.
The authors are indebted to Iman Sudarwo, Minister of State for Research and Technology, and the entire IPTEKnet team at BPPT (Agency for the Assessment of Applications of Technology) for their cooperation in this project, as well as Darias Manns and Robert Schware of the World Bank. Lerzan Aksoy of George Mason University's Graduate Business Institute and Steven Huter of the Network Startup Resource Center (NSRC) at the University of Oregon also provided valuable assistance.