iassist quarterly - 1

Developing A Social

Science Microcomputer
Local Area Network:
Its Uses and Problems

features.The microcomputer local area network

is more than a technological fad, it is

convenient and elegant way to add additional

function to exisiting equipment for a moderate

price. With proper planning and managment,

local area networks' uses far exceed their

problems. This paper will briefly discuss the

basic definition of a local area network and

review some of the points to investigate when

choosing one. It will then cover the problems

and solutions one might encountered in

constructing and running a social science local

area network.

by Fred Nick

Center for Social Science Computation and

Research

University of Washington

Rgure#1

Abstract

A local area network offers a social science

microcomputer user many nice advantages but it

can also create a host of problems for the

individuals in charge of network administration.

As social science microcomputer use continues

to increase, local area networks will offer social

scientists welcome communication, file storage

and peripheral equipment sharing capabilities.

For the network administrators problems arise

in: initial funding for the additional equipment

needed, the definiiion of which network to

install, the added security problems in running a

network, additional responsibilities for storage

backup, the establishmeni of access policies, the

day-to-day maintenance of the network and its

software, and providing the additional help

needed by the users of the system so that they

can take advantage of the new network

Numb^of PCson Multlus^

PC Systems (thousands)

40000

30000

20000

10000

1985
year

1990

I

Tolal number of PCs In uje (thouwnds)

I

N urn b er of PC3 on Multiuser PC sysletnd

(thousands)

Fall 1986



8 - iassist quarterly

What is a Microcomputer Local Area Network?

A brief definition of a miaocomputer local area

network (LAN) might be as follows: the

equipment and software necessary to connect

microcomputers to each other so that they can

share files and peripheral equipment while

establishing inter-microcomputer communication.

Some LANs also enhance the operating

characteristics of connected PCs (personal

computers). A network usually includes a

workstation which functions as a file or disk

server, containing files stored by users and

administrators of the network. This server is

connected to the other network workstations by

a cable. There may be more than one server

on a LAN. and often there are many printers

and other peripheral equipment attached to the

network.

figure ;f 2

Projected installations of

shared processors & network

file servers (thousands)

3000

2500

2000

1500

1000

500

i^M

m

--

i

I

i

lil^ J .

11

1985 y^^ 1990

Industry analysts predict that the number of

miaocomputer LANs will be 4 to 6 times

greater in 1990 thjm in 1985 (see Figures

1-3)'

In 1985, 26,500 networks were purchased; it is

projected that there will be 121,270 networks

sold in 1990'

figure #3

Anttcipated grovvth or PC

network shipments

HOOOD

120000

100000

8OG0O

60000

40000

20000

1984 85 86 87 88 89 1990

year

'Lilv, Susan Vendors ready network versions of
sojware. PC WEEK 2(25): 102, June 25. 1985.

(Graphics included as figures 1-2 above)

Lilv, Susan "IBM token ring opens doors for

independent net vendors" PC WEEK 2(51):113,

December 24-31, 1985. (Graphics included as

figure 3 above)

Fall 1986



iassist quarterly - 9

Initial Enviroiunent

The first stage in establishing a local area

network is careful planning. Networks can be

ver>' expensive, and have a large number of

differing capabilities; a bad choice could be

costly. The first step in planning is to examine

the environment in which the network will be

placed. What equipment is presently being

used? How much experience with

microcomputers does the user community' have?

What fimctions are needed, and what

capabilities are patrons likely to anticipate

having available. A primary' complaint about

LANs is that they have not performed as the

users had anticipated. Unfulfilled expectations

can make a system appear to be a failure when

it had merely not been designed to do what

people requested of it

As examples, I will draw on the installation of

a LAN in my organization, a computer support

group which helps soda! scientists to use

computers. It has existed for 13 years, and has

traditionally been mainframe oriented. It is

located in a state-funded university, and derives

its entire budget from state monies. It has no

mechanism for recapturing expenditures through

charges. All services are free to users of the

facility. All users of the system were familiar

with mainframe-style interactions and

capabilities. We had purchased an Apple 11+

five years earlier, but ii had proven a failure.

Not enough software was available at the time;

the operating system was too simple, and the

functions demanded by the mainframe users

who experimented with the microcomputer were

not available. Thai purchase was ahead of its

time.

Three years ago we again became interested in

introducing microcomputers to our patron

community. A new generation of computers

became available, major software packages were

creating renewed interest, and microcomputer

word processing was becoming popular. Also,

several major computer companies were starting

to give equipment grants to universities. We
therefore decided to try to develop a

miCTOcomputer LAN. The Apple 11+ experience

influenced the decision-making process when we
began to plan for a miaocomputer LAN. Our
user would demand certain 'mainframe'

characteristics. There must be 'personal storage'.

Patrons were very imeasy about having files that

any user could read, manipulate or destroy.

They were accustomed to security precautions

such as passwords and access modes. If our

LAN was to include shared storage, on the

same disk, then these features would be

requested. Since the users had mainly

mainframe experience, it was essential that we
provide access to the mainframe so that files

and programs could be moved freely to and

from the LAN. We decided that since our

organization had many diverse groups of

potential users, we should choose one group and
develop the LAN with those individuals in

mind. Because we had little chance of

developing a large LAN, we decided that a

small LAN would have the most far-reaching

exposure if it were dedicated to faculty

education and instructional development

Faculty and teaching assistants interested in

gaining skills on miaocomputers or developing

microcomputer courseware were allowed access

to the LAN. Our staff developed free

non-credit courses to educate novice users.

Several other campus microcomputer facilities

had developed severe managmeni problems

when word processing was allowed on them.

The microcomputers had quickly become

24-hour-a-day word processing stations to the

exclusion of all other uses. We therefore did

not allow 'production' word processing.

Development of word processing skills was

allowed if noi encouraged, however.

Having identified what you want, whom it will

serve and a few of the operational details, how
do you gel started?

fall 1986



10 - iassist quarterly

The first two tasks are obtaining financing and

finding an appropriate space for the LAN. If

you can charge for services, you should

experiment with possible funding mechanisms to

determine if the costs of developing and

supporting the LAN can be recovered. LANs
seldom have a convenient method of keeping

track of useage. Charging by the hour, or a

monthly fee, are possible alternatives. It is

difficult to differentiate between "heavy" users

and "light" users other than by the clock time

they spend using the equipment Even

consumables, such as the paper used in printing,

is difficult to monitor on some systems; other

systems record the number of lines printed.

Some granting agencies may be appropriate

sources of funds to finance the LAN. Detailed

advance planning will help organize and collect

the information needed for gram applications.

At many institutions of higher learning, physical

space may be more difficult to obtain than

money. Each LAN microcomputer requires

approximately 16 to 20 square feet of space,

when placed on a table with a chair in front of

it. Extra power outlets will be needed, and it is

important to remember that microcomputers can

generate lots of heat (as do their human users).

Proper electrical connections, ventilation and

heating must be considered. Also security must

be considered when looking for space.

Microcomputers are very popular booty for

thieves. Consideration must be given to which

fioor of the building the room is on, where the

windows are, how the room is locked, who has

access to the room now, and if it can be

rekeyed. Careful planning can prevent many
managment problems later.

In developing our LAN both space and funding

were critical points. Our financial situation

could not have been worse when we decided to

pursue obtaining a LAN. Our budget came

directly from the state legislature. The state

was in a financial crisis. Our operating budget

had just been cut by 20%. There were no

capitol funds available for equipment purchases.

The Dean considered the project an experiment,

and there was a considerable amount of

prejudice against microcomputers to be

overcome. Undaunted, we decided to proceed.

We went directly to computer manufacturers,

asking them for gifts or loans. To our delight

and surprise, the IBM Corporation agreed to

loan us 5 IBM XTs, 5 printers and some

software to get things started. These were to be

loaned to us for 6 months, but in the end we
had kept them for one and one-half years.

After 6 months of our having used the

borrowed XTs, the budget situation had

improved greatly. Use of the borrowed

equipment had also quelled many criticisms

voiced earlier. At that time we were given

$40,000 by the University to purchase a LAN.
Many projects had been started on the

borrowed equipment, which had laid a solid

foundation for the request for a LAN.

Space was donated by another department,

electrical outlets were nimierous, and the power

was not polluted by other equipment on the

circuit (Having other electrical equipment on

the same circuit could cause voltage fiuctuations

which might affect the LAN equipment). The

room had no windows, and the door was

rekeyed so that even the janitors did not have

access. We did not plan for ventilation,

however. The room would become terribly

warm with just a few individuals in it Since

there was no venlilalion system, nor windows,

users would open the door to allow the room to

cool off, undermining our security and allowing

access by unauthorized individuals. We
eventually had to trade rooms to remedy the

ventilation problem. We had no funds with

which to have proper ventilation installed.

fall 1986



iassist quarterly - 11

Which Network to Buy?

Choosing a networl: can be very confusing.

There is a variety of technologies, hardware

requirements and functional difTerences.

Networking technology has its own set of "buzz

words", or jargon, which is highly complex.

There are many topologies or workstation

layouts. Most of the literature is difficult to

read without first having learned the definitions

of the most corrunonly used phrases. The best

approach in choosing a network is to avoid

becoming too involved in comparing all the

performance figures quoted in the literature.

First, define how the network will be used, and

then determine which network characteristics are

important If large files stored on a central disk

are going to be used, transfer speed may be an

important criterion. If users are to be able to

converse from one station to another, check to

see that the nodes on the network can address

each other and send messages directly to the

screen of another node. Electronic mail is an

attractive method of communication which is not

reliant upon the recipient being at his desk

when you want to communicate with him.

Electronic mail can store and organize

conespondenses, forward messages to others,

and send replies along with the original

message.

Cost is an important criterion. Networking

hardware and software can be very expensive

and is seldom inexpensive. If you are building

a network of exisitng PCs, it may be possible to

use some existing hardware to provide the

comunicalion connections. Low speed LANs can

use standard serial cards as communication

hardware. Networks can cost more than a

$1000.00 per station to install. Always try to

balance cost and function. One network may
cost a little more than another, but offer far

superior performance or stability.

LAN technologies have length limitations

between workstations, and usually have

limitations on how far apart the first and last

node can be. Limitations on distance on a low

speed LAN will vary according to the speed at

which the equipment is seL If the PCs are in

different buildings, certain networks may not

work at all. It is suprising how far apart two

terminals can be if the wiring has to be run all

over the building. Distances must be measured

accurately before the hardware is bought Since

different networks use different hardware, the

life of a network can be increased by choosing

one which allows one to change from one brand

of software to another without changing the

hardware. Ethernet cards are used by several

LANs as well as serial cards. The reputation of

the company is another good indicator in

making the choice of network. If nothing else,

having a popular network insures that one may
be able to obtain help from another local user.

What might you have to give up if you already

have PCs and wish to install a network? Some
networks alter the operating system. When the

next version of the operating system is released,

it may not be usable. PCs may behave

differently, thereby causing present users some
problems. The machine which becomes the

server may no longer be usable as a regular PC.

What might you gain by installing a network?

Enhanced PC capabilities are one possible

benefit Some networks make it appear as if

the PC has more hard disks than it realh has.

You may gain electronic mail features. Access

to printers not directly connected to the PC is a

common benefit of LANs. Files stored on the

server may be protected by security access and

logins with passwords. The abilit\ to isolate

individuals' files and label printouts is another

nice feature.

Fall 1986



12 - iassist quarterly

Choosing a Network: A Case Study

My organization had 6 months experience using

IBM XTs when we made our decisions as to

what equipment to buy and which network to

install. The day the IBM XTs had arrived no

one but myself was suppose to know of their

arrival. The equipment was delivered in large

cases stamped "IBM" all over the sides. By the

end of the first day, I had received 20 requests

for keys to the room where the XTs were

placed. These 20 people were all interested in

doing free word processing: a clear hint that

word processing would a problem. Other

management problems quickly emerged. Floppy

disks were being stolen. Faculty accidently

reformatted the hard disks on a weekly basis.

Bootlegged software appeared on the hard disks

even though we had instituted strict rules

against its use. Many policies had to be

hurriedly created in order to solve these

problems. We decided to buy IBM
microcomputers and connect them with a 3COM
LAN. The LAN was viewed as a potential

solution to many of the problems mentioned

earlier. We had several users interested in

using social survey data on the network, and

therefore decided to try to obtain as high a

transfer speed as possible. Because of

"ownership" problems experienced with more

than one person using the hard disks, we

needed a network which would have access

passwords for each user, and provide exclusive

control over certain administrative functions.

We also wanted to enhance the functions of the

network PCs, since we were buying expensive

network hardware, rather than extra disk drives

and printers. We purchased IBM single drive

PCs with 512k of memory, color monitors, 8087

math coprocessors, and a multifunction card

which supplied each machine with a serial porl,

parallel port, and clock. The peripherals

included a 6 pen table plotter, an ink jei

plotter, a digitizer, and a dot matrix printer.

We decided to buy a 3COM network because it

satified most of our criteria. It included a

$1000 ethemet board and software for each PC
(these now cost about $600). In addition, we

had to buy thin ethemet cable and three pieces

of software: file server software ($750), mail

software ($750), and print server software

($500). 3COM met our criteria in the following

ways:

a. It is very fasL It transfers information at

10 million bits per second. This was

much faster than most other networks.

b. It offers excellent security. Each user has

an account and a password. Files

belonging to one user can only be access

or altered by another user if the owner

gives permission. The adminisiative

software and system library are protected

from accidental erasure and the

administrative menus are available only

after a control password is repeated.

c. The network dramatically enhances the

performance of each PC. Each individual

machine has only one disk drive, but

when logged into the network, performs

as if it had 4 additional hard disks. Each

hard disk is emulated by the network and

contains only information stored there by

the owner. (Each user creates his own
"virtual" disks.)

d. The network commands are separate from

the operating system, not patches to it.

They are easy to use and there is a good

help feature.

e. The network automatically captures all

prim requests, including screen dumps,

and directs them to the system primer

where they are printed with a banner

page on the front displaying the owners

id.

r The electronic mail facility is elegant and

Fall J 986



iassist quarterly - 13

has many features.

Users cannot control or contact another

user's tenninal except through electronic

mail. We viewed this capability as a

problem and were reheved that it was not

available on the 3COM network.

At the time that we made our decision,

3COM had the largest number of working

networks in service, and after talking to

several satisfied owners we felt that the

network was reliable and that there were

many possible sources of information and

assistance.

The installation of the network was

straight forward and the installion

instructions were ver\' clear.

What were the negative aspects of 3Com? The
following:

a. It is very expensive. We spent twice as

much on 3Com as we would have spent

on other networks.

b. 3COM allows any user to generate

another account for himself Only two

individuals have discovered this feature,

which is apparently an oversight in the

networking administrative software.

c. 3COM allows any user to delete another

user's account but only if the account has

no stored files. In practice, this is not as

much of a problem as it may seem.

Every user on our system has a mailbox

file, and therefore every user has at least

one file.

The Uses of a Social Science Microcomputer

LAN

The social scientist can take advantage of a

LAN for both instructional and research

applications. On the research side, a LAN
provides the user with a larger, embellished

machine on which to perform analyses. Access

to 4 virtual hard disks allows versatility in the

organization and storage of text and data. The
speed of these virtual disks can increase

performance when using software that accesses

multiple files simultaneously. (There is no need

for one of these files to be on a floppy disk.)

In addition, software can be stored on the

system disk and many users can use these

libraries simultaneously. The LAN workstation

functions like a slow minicomputer rather than

a single microcomputer. Many of the

frustrations involved in using a microcomputer

are alleviated upon installation of the LAN.
The communication capability is a hidden

blessing. Leaving messages for other members
of a research team becomes an addictive, useful

tool. On the instructional side, the LAN allows

a single copy of assignments and data to be

stored, instead of one on each machine. Again,

mail allows faculty and student interaction

outside classroom or office hours. The software

available for microcomputers is often easier and

more fun to use than mini- or mainframe

software. LAN software is almost as elegant as

mainframe software. It is also important to

remember that microcomputers are now firmly

entrenched in the business community. The
microcomputer exposure that the students

receive pays off not only in the classroom but

also later when they are in the job market

Many interesting applications and programs were

developed on our insiruclionai LAN. The
Departments of Politicai Science, Public Affairs,

Geography, Demography, Computer

Cartography, Economics, Sociology, Social Work
and Psychology all took advantage of the LAN.

Fall 1986



14 - iassist quarterly

In Political Science, a simulation was constructed

to illustrate how community political pressure

causes the development of a 911 response

system. Economics constructed

telecommunication software to enable the LAN
miCTOcomputers to act as elegant graphics

terminals interacting with large economic time

series databases stored on the mainframe. In

Demography, population simulations were

developed which graphically displayed

population pyramids. Geography/Cartography

developed a continous shading Choroplethic

mapping program. We trained 5 Social Work
faculty members to leam Basic so that they

could program the IBMs they had at home.

There are many other experiments and many
faculty have increased their miaocomputer
skills.

Problems in Using a Social Science

Microcomputer LAN

There are two major classes of problems with a

LAN: problems for the users and problems for

the administrators. Novice users are in need of

extra support to help them begin. Purchasing

good computer tutorials for the software is a

wise investment, as is creating locally written

documentation which highlights what options to

leam first and how lo use the help features of

each program. The major problem we
encountered was that the operating system for

IBM PCs is terse and quite frightening to many
novices. Our solution was to obtain a gram to

hire a programmer, for a summer, to write a

menu driven control program for the LAN.
Each user merely inserts a disk and turns on

the microcomputer. From that poini on, a

series of menus asks the user which programs

he wishes to use. After he is finished using the

selected program, the menu system again takes

control, cleans everything up, and logs the user

off the network. It is an elegant piece of

software which works not only on a 3COM
LAN but also on any single IBM XT computer.

The software also allows instructors to create

class menus which providing their student with

access to only that software necessary to

complete their class assignments.

There were more administrative problems than

user problems. The LAN demands a great deal

more adminstrative attention than I had

envisioned. There are constantly updates to

software to be installed. To install software

requires taking the lime to become familiar with

the program, its organization and its options.

The installer is the first source for inquires

about which version of the software is available,

and is this or that option available. Backups

must be made of the system disks on a regular

basis. If the hard disk needs to be reformatted,

it may be necessary to invest several days of

work to rebuild it, and confirm that it is now
working properly. Hard disk management is

also time consuming. Reorganizing storage on

the disk will be mandatory several times during

the first few months of operation. Just to

decide what software to buy can use days of

time. There is no repair service that can be

called in if the network stops working. If you

are successful in developing a small LAN, you

will probably immediately be asked to build a

deliver>' LAN large enough to handle large

classes. Everyone will demand use of the

equipment for word processing. If the resources

are available, there is little reason not to

provide this service, but if resources are limited,

one must beware of setting an unfortunate

precidenL Everyone will want a key to the

facility. But, one unlocked door can result in

the loss of your entire LAN to thieves.

Insurance on the equipment is a good

investment, if the budget allows. We had noi

anticipated the legal problems of software

licenses and software theft It is the

administrator of the LAN who is primarily

responsible for upholding software contracts, as

well as making sure that state, local and

fall 1986



iassist quarterly - 15

university laws concering software theft are

upheld. When reading software contracts,

beware of clauses prohibiting use of the

software on a LAN. Several major software

producers have such stipulations. If the license

states that the software can only be used on

one machine, indicate to the users which"

machine they must use to run that software. If

possible, use of that software on more than one

machine at a time shoiild be hindered. When
buying software that will be used by several

simultaneous users, one should buy the legal

number of copies or inquire about LAN
versions of the software. The laws which

concern software theft should be posted in the

room containing the LAN. It is important to

ensure that siafT uphold these laws and that

they project the right attitude about theft to the

users of the system. Do not turn your back on

illegal copying. If someone is caught, it is the

administrator who will be indicated as legally

responsible for the protection of the software

license.

c. Plan before you buy.

carefully.

Identify needs

Plan to avoid obsolecence. Try to

purchase from companies which will help

the network evolve and expand, especially

as national standards for network

structures appear.

Security should always be a primary

consideration in developing budgets,

buying equipment, and choosing space.

Beware of setting bad precedents.

Ask the network salesperson for a list of

software and/or hardware that not work

on their LAN.

Having Installed a Successful LAN, What Next?

Problems Versus Uses

Trouble?

Is a LAN Worth the

Definitely! With proper plarming. the problems

can be reduced lo a minimum and the value of

the LAN to the users increased. Some steps

which will streamline the installation of the

LAN:

a. Hire a LAN manager. If the funding is

available, this will have a dramatic

positive effect on the installation of your

LAN.

A bigger LAN!! Because of the ease of use

and friendliness of microcomputers and the

added utility of the LAN, users will demand
faster machines and more software. Also, if you

do not have a closed community of users, your

LAN converts will encourage others to use the

system. Use in an instructional setting can grow

exponentially with just a few moderately sized

classes moving their work onto the LAN. My
group is installing a 22 IBM AT LAN using the

IBM ring network this summer. Again a grant

from IBM purchased the equipment. The

experiment starts over again, but on a larger

scale!!

b. Take advantage of student employees.

Especially in the beginning, the LAN will

involve a number of small tasks which

are time consuming. Student employees

offer talented help at an affordable price.

Fall J 986



16 - iassist quarterly

References:

Bergin, Thomas J. Establishing a computer

laboratory to support the teaching

of the social sciences. Social

science microcomputer review

3(1): 14-27, spring 1985.

Chapman, Dale. Campus-wide networks: three

state-of-the-art demonstration

projects. T.H.E journal

13(9):66-70, May 1986.

Derfler. FranL Recent changes on the PC
LANscape complicate buyers'

choices. PC WEEK sec. 1

l(43):69-82. October 29, 1985.

Jenkins, Avery. With icons and menus, LAN
operating systems meet the

demands of both file servers and

network users. PC WEEK 3{7):61

ff. February 18, 1986.

Lily. Susan. Vendors ready network versions of

software. PC WEEK 2(25): 102,

June 25, 1985.

Lily, Susan. IBM token ring opens doors for

independent net vendors. PC
WEEK 2(51): 113, December 24-31

1985.

Rauch-Hindin, Wendy. Universities are setting

trends in data commimications nets,

pp. 202-212 in: Davis, George R.

(ed.) The local network Handbook.

New York. N.Y.:McGraw-Hill.

1982.

Thurber, Kenneth J. and Freeman, Harvey A.

The many faces of local

networking, pp. 222-230 in: Davis,

George R. (ed.) The local

network handbook. New York,

N.Y.:McGraw-Hill, 1982.

Way, Dale. Build a local network on proven

software. Davis, George pp.

230-233 in: Davis, George R.

(ed.)The local network handbook.

New York, N.Y.:McGraw-Hill.

1982.

fall 1986