lA^biiiT Newsletter, Vol. ^, No. i (iJummer lyyttj NtTWOKKS ANU HETWOHKING Thomas Win. Madron Western Kentucky University AbSTHACT Tne purpose ol this paper is to provide an overview or computer net- worKS and tneir associated communications systems. In developing that overview we will first define the idea ot a "network", then describe com- munications systems including commentary on equipment and costs. Having aealt with the role ol communications we will then turn to computer net- works per se and review the contributions networks can make to computing. The discussion of computer networks will deal with shared hardware resources, shared software, types of systems, and network cont igurations. The general discussion ol networks will be followed by a description of a developing network: the Kentucky tducationai Computer Network IKECNtT;. The paper will conclude with a comment on some of the opportunities pro- vided through the use of networks. In the following pages we will attempt to explore some of the characteristics, problems, and iwl KUUUCTIUN opportunities provided through the establishment and use of computer Ihe title of this paper, "Net- networks. Because computer net- works and Networking", may not pro- works are dependent on communica- vide the reader with the topic as tions networks, several attributes sell-evidently as it might seem. of communications systems--as those The terra "network", as used by systems apply to computers--wlll those of us Who use computers, may first be described. Ihe core ot refer to communications networks, the papei centers on a discussion computer networks, or both. The of computer networks and their topic of this paper is directed characteristics followed by a dls- primarily toward computer networks, cussion of various configurations although It will become evident networks might take. Ihe more gen- that we will have need to speak of eral discussion of networks will be communications networks as well. made concrete by describing one operating network--the Kentucky What are computer networks? One Educational Computer Network (KEC- commoniy used definition oi a com- NEl J--wnich is still m the process puter network is the following: of growth. Unally, we will One or more computers accessed by attempt to bring theory and prac- users via some communications net- tise together m a summary discus- work. The computers may consist of sion ol why we should use networks main-site, or nost computers, and what applicatlons--especiaily and/or remote computing systems. in higher educatlon--are possible. Generally it may be said that com- munications networks can exist without computers, but computer networks cannot exist without com- munications networks. - bo - iAbiJi5J New3ieti,er, Vol. d. No. i Ibumnier ly^tt) LUMMUHiCAl lUNla Nhl WUKK,b Danowidtn. 10 transmit data over a telephone line, whether puDlic or A communications networK con- private, that data must be manipu- sists of a communications medium iated electronically so that it and computers or other devices used fits into some segment of the fre- in the control ol the communica- quencies into which a channel may tions process itself. The most be divided. This manipulation is pervasive communications system accomplished by a device called o available to most ot us lor voice- modem. Ine speed of the line--tne grade exchange is the telephone rate at which data are transmit- system. Ihe telephone system has, ted--are measured in "bits per sec- 01 course, been witn us lor a con- ond", sometimes called baud rates, siderably longer period of time in the period from about ly^U to than have coiiiputers. While the about I^Y^ there was over a 4uu telephone is the basis lor the most percent increase in he normal baud common communications networK, rate at which data could be tran- there ,iTQ other such systems linKed smitted over conventional telephone by land-lines, microwave or other lines (,cf. Martin, lyCU, p. Y, for radio transmissions, and, most the earlier rates), recently, lasers. because most communications systems predated the need for data transmission, there has been a tendency to provide Communications tquipment means oy wnicn digital data can coexist with voice messages. in While we cannot delve very far order to control the communications into tne establishment and control process it nas been necessary to ot communications networks in this provide instruments capable of pro- paper, it would oe well to explore viding the proper level of service a lew o^ the devices needed in the lor data handling. transmission of data. Among those having importance for our purposes In a nor.iial telephone conversa- are modems, and their extensions, tion, two or more people speaK to multiplexors and concentrators, one another over lines connected by means oi public exchanges. Jhese Ihere is a relatively narrow lines are called "public" or range of frequencies which may "switched" lines and can be used travel without much distortion over for communicating data as well as telephone circuits. ihe human voice coinmunications. Ihe alterna- voice has a middle frequency ol tive to public lines--althougn about ibUU hertz. if we translate still provided by telephone compa- binary data into a modulated fre- nie3--are "private" or "leased" quency ol about ItiUO hertz, then lines which are connected perman- data too can be transmitted without ently or semi-permanently in a data great distortion. A device called communications system. Hegardless a modulator is used when sending of the kind of line used, the sig- data in order to achieve the neces- nai carrying capacity of communica- sary modulation ot the data. At tions lines is usually described in the other end, a demodulator is terms of frequencies those lines used to change the modulated car- wiii carry. Ihe range of frequen- rier back co normal binary data a cies IS called the bandwidth ot the computer can use and understand, channel and the quantity ot data Normally these units--tne modulator that can be communicated over a and demod ulator--are combined into channel is proportional to tne iA5bii>T Newsletter, Vol. d, No. i lijummer ly/tt) a single unit with the abbreviated is consumed by a chunk of one ot name ot modem. Modems are made by the signals to oe sent, independent vendors as well as by telephone companies. Ma bell often While multiplexing equipment is calls modems "data sets." Although relatively inexpensive, and while It IS outside the scope ot this multiplexing may be a convenient paper, it should be noted that method ot using one high speed line there are at least three dilferent tor a number ot lower speed tasks, types of modulation. Moreover, there are some occasions where mul- data processing equipment may be tiplexing does not work well. Mul- directly wired to the telephone tiplexing is limited by the total line or may be connected via an bandwidth available thus providing acoustic coupler when dial-up is a limit on the number ol channels the normal mode tor accessing a into which a given link may be sub- target computer. divided. Une way ot getting around the limitations is to use a "hoid- une ot the problems with tele- and-torward concentrator" which is pnone lines is that they are rela- nothing more than a small computer tively expensive. Une way ot cut- which receives messages trom a var- ting costs, especially when using lety of terminals, stores the data relatively low speed communication in some memory (butters;, then equipment, is to employ multiplex- tires the signals up the line to ing . Ihrough the use ol multiplex- another computer, ihe advantage of ing It is possible to send signals a concentrator is that many ditter- Irora more than one terminal over ent kinds of entry devices may be the same line at the same time. In attached to it, all running at dit- a multiplexed system, two or more tering baud rates, and the capacity signals are combined and transmit- is limited only by the capabilities ted over a line with an appropriate ot the concentrator itself. Unlike bandwidth. Ihe terra "multiplexing" multiplexors, however, which are then, means the use of one facility transparent to the end-user, con- to transmit in parallel several centrators may cause delays in tur- ditterent channels oi data over a naround and, being more complex, single communications linK. are more subject to tailures of various kinds. Judgments as to une ot two multiplexing methods which kind of approach to use are normally used: f requency-divi- should be governed largely by sion multiplexing or time-aivision cost/benefit considerations in tne multiplexing. t-requency-division overall design ot the network, multiplexing requires a modem or data set that provides several groups of channels over one line, dividing the total bandwidth of the Cost Considerations line into channels of some smaller frequency range. 5uch a device as Martin has noted Ciy^u, p. pertorms the normal t unctions of a i;, the "tacts about the transinis- modem while carrying out the muli- sion links which most concern a plexing. By way ol contrast, sig- systems analyst are the cost of the nals may be sent in a round-robin links and the rate at which data fashion so that only one signal may be sent over them." it is occupies the channel or line at any becoming clear that the cost ot one time. ine time available is commuications is an increasingly divided into small slots and each important tactor in the cost ot iAijbibT Newsletter, Vol. d. No. i Cbummer lyYO; computing and as large computer Shared Hardware Kesources networks expand, and as we become more dependent on networks, commu- Certainly one ol the early moti- nications costs will expand at a vating tactors Deiiind tne estaoi- signilicant rate. It nas been ishment ol' computer networks was estimated tliat by lycit) ninety out the fact that tnrough networks many ol every one hundred computing doi- users could use expensive computer lars will be spent for data commu- systems, thus spreading the cost ol nications rather than I'or the data such systems over a much larger processing itself Ci-erreira and user base. Very often the total Niiles, lyYbJ. As a result of the number of users on a large network cost considerations and potential can support equipment on a scale no expenditures in the future, systems individual user or user institution analysts and organizational manag- could manage alone. "why both net- ers will have to become more cogni- works and stand alone computers", zant of the demands for allocating one might ask, "when mini and micro resources for telecommunications. computers are getting to be so A byproduct of the projections inexpensive?" While it is true noted is that manufacturers av% that some micro computers cost lit- increasingly turning to the produc- tie more tnan expensive terminals, tion of communications hardware and and while the capabilities of such software (.Xotaro, lyYo;. in the computers are great, they remain end, however, the objective is not inappropriate in situations where only to move the data, but to pro- the same data base must be accessed cess It. Consequently, we should by many different users, where consider the computer network which files are large and must be IS supported by the communications accessed rapidly, and/or where user system just described. programs require very large amounts of memory to do the task needed. It may be helpful to look at each one of these points in turn. tUMHUTKH HETWUHKS Two of the three points just Why should computers--and there- noted deal with capabilities lore end users--be organized into regarding files and file struc- networks in the lirst place? With tures. Although the cost of compu- computer hardware becoming less ters is falling rather rapidly, the expensive and at the same time more cost of large scale peripheral dev- sophisticated , the answer to the ices--devices which can be used to foregoing question is likely to be permanently store large quantities somewhat diiferent in the latter of data--are not decreasing in cost ly/U's than It was earlier in the m any substantial fasliion. More- Uecdde. '.lo j iart^e extent the over, even snould large-scale peri- reiiidinder of the paper will be pheral devices fall in cost, the devoted to answering the question lower price for hardware does not "why networks?". Before going necessarily justify using such dev- lurther, however, it will be help- ices to keep multiple copies of the ful to explore some of the outcomes same data base. If, in a given and characteristics of networks, application, several small compu- then turn to a description of the ters can legitimately be used to various configurations networks may replace one large computer, then take. the decision should be made on the basis of the comparative costs of - by - lAiii>i3T Newsletter, Vol. d. No. i (.Summer ly/a; tne two approaches when aii elements ol' the two are consiaered as complete systems. Notwithstanding the comment con- cerning tne cost ol large-scale peripheral devices, we should note that devices such as large diSK dr ives--wnich are very costly iteins--are likely to be replaced in the relatively near future Py elec- tronic memory systems (such as the bubble memory), thus eradicating the arguments just presented, similarly the question ol the amount ol main memory needed lor user programs will become less pro- blematic as the cost ol' memory is reduced and as micro computers are designed to address ever larger Chunks of memory. None ol the solutions or substitutions lor net- works are nelptul, however, when one IS dealing with a large data- base which must be used by various users located in diilerent places. Shared Software Hesources While the ware capabi mini, and mic ing blurred , necessary for operation of any size is b portion of t ning such sys taries on t development o have suggeste several years more and mor languages mo natural langu programmer wi standing tha lact, however see signific personnel tim programs whic user to u differences lities among ro computers the personn the program computer sy ecoming a la he total cost terns. Variou he direction f computer f d that over , as comput e to be prog re nearly r ages, the ro 11 change, t fact, or , we will CO ant contribu e to the pro h will allow se computers in hard- large , is becom- el costs ming and stems of rger pro- of run- s coramen- of the acil ities the next ers come rammed in esembl ing le of the Notwi th- alleged ntinue to tions of vision 01 the end in a reason As a r sidera sidera cerns about the ha ciosel shared sectio provid contin works availa previo ing ex discus comput Idea would notion lashio ably s esult tions , tions about the CO rdware y rela data n. t es arg ued us can ble to usiy p ample sed in er CO of com not b 01 ne n . traig of s the will hard St of l' ted noted ach urn en t e of also use resen Ola some mere puter e tea twork htfo uch refo Shi ware the his to t in of t s in net ma rs w t . n det ncin ized sibl s of rward perso re , ft f to peop same he p the hese fav works ke hich une dded ail b con wi som manner . nnel con- cost con- rom con- concerns le to run issue IS roblem of foregoing concerns or of the Net- resources were not interest- resource , elow, is The very ferencing thout the e form or lypes of Systems Computer networks may be struc- tured to support remote batch, or timesharing, or some combination of the two. Alternative names for batch systems are "remote job entry IKJE)" and/or "remote batch termi- nal IKBi;" systems; the most cur- rent buzz-word is "KtsJ". liraesnar- ing systems are often called "interactive" systems. Kbl systems are tnose which operate essentially with card input (or its equivalents in the form of floppy disks, on- line job entry, etc.;. Jobs are placed in a job stream, given a priority for the order of execu- tion, and a job in its proper turn IS executed and returned to the user. Unce the job is submitted it is out of the hands of the user and dependent on the characteris- tics of the operating system under which the job is being run. In contrast to a batch approach to networking is the use ol timeshar- ing or interactive systems where the user sits at a iAii^iST Newsletter^ Vol. e' , No. i (Summer lyVd) typewriter-styled terminal and the oversimplification ot networks, we computer and tne user interact with will review four characteristic one another in real time. in such conl igurations : point-to-point; systems the user always has control multipoint; centralized; and hier- over what is happening to his/her archical, joD and may make immediate res- ponses to any problem which arises. Ut ten hyorid systems are developed such as IBM's conversational job Pomt-to-Hoint entry system CCJ3; or UtC ' s approach to tUKJKAN and tlUbUL on A point-to-point network is, tneir large systems. In Doth without douDt, the simplest kind of approaches coding for programs is network for it consists ot a compu- entered interactively through a ter, a telephone line, and one ter- text-editor, but the program, when rainal at the other end ot the tele- tinished, is actually submitted as phone line. Ihe terminal can be a batch job. Both approaches to either an KBT or interactive. This networking have positive and nega- simplest of systems is depicted m tive teatures, but what will be Figure 1. Many networks begin as said m subsequent sections ot this point-to-point systems and gradu- paper apply equally to both types ally develop into more complex ot systems. entities. Network Uont igurations Multi-point Networks There are a variety of ways in Multipoint networks constitute a which networks might be configured straightt orward extension ot ana many (perhaps most; networks point-to-point systems in that are in a constant state of change instead of a single remote station, and growth. If the computer net- there are multiple remote stations, work consists of only a main-site Those remote stations may be either or host computer which does all KBT's or interactive or a combina- data processing from one or more tion ot both. Ihe remote stations remotes, it is a centralized net- may be connected via independent work. It there are remote compu- communications lines to the compu- ters processing jobs for end-users, ter or may be multiplexed over a as well as a main-site computer single line. Such a system is (which IS itselt optional;, then we illustrated in l-igure t^ . in either may have the beginnings of a dis- a point-to-point system or a multi- tributed network. A distributed point system, the characteristics network can be either centralized ot the remote work stations are a or hierarchical in lorm, but a net- function of the work to be accom- work which does not involve distri- plished at the remote site, buted processing can only be cen- tralized since all data processing is done on a main-site computer. It IS possible tor a single commu- Centralized Networks nications system to provide commu- nications services tor two or more as noted above, a centralized concurrently operating centralized network is one in which primary computer networks. Although the computing is accomplished at a sin- comments which follow constitute an iAb:5i5l Newsletter, Vol. d, No. i tiiummer lyYB) Hierarchical Networks Mgure I: Point-to-Point Network gie site witn al feeding into that a system is thou network with e entering the cen single communicat m Mgure f,. bot a star networK ar terns in that they a single, cent lypically, howev network may not processing capab star network may ters out at the e cations lines. 1 ter which suppo networK might , i into a star netwo 1 remote sta site. Often ght of as a ach remote tral system ions line as h a multipoin e centralized are controll ralized comp er , a multi have distri ill ties whi nave other c nd of its com n tact, the c rts a multi tsell , De 1 rk . tions such star site via a noted t and sys- ed by uter . point buted le a ompu- m u n 1 - ompu- point inKed A hierarchical network repre- sents a tully distributed network in which computers feed into compu- ters which in turn feed into compu- ters. The computers used for remote devices may have independent processing capabilities and draw upon the resources at higher or lower levels as intormatic i or other resources are required. Such a network is pictured in Kigure 4. A hierarchical network is not the only kind of distributed network but It IS a completely distributed network. M Figure ^: Multipoint Network iAbbibI Newsletter, vol. d. No. i Ciiummer ly/B) Hgure J Centralized or 5tar Network At se we have distribu two con opposite although can d i thought com put in gested b Lusa ny coinputin part of ing of d tne pia nates an taming work." networks simple veral point spoken of ted proces cepts in n ends of such IS no str ibuted of simply g. One d y Data 1UU YbJ states g places the pre- a ata, and a ces wnere d is used . central con In today' s which star star system in centr sing a etwork some t the proce as dec efinit as r that d "a s nd pos ccess the da . . w trol o comput t as oft this paper alized and s if the s were at continuum , case. Nor ssing be entralized ion, sug- eported by Istr ibuted ubstantiai t-process- to data at ta origi- hile main- f the net- ing world, relatively en become distributed systems without cons- cious design on the part of systems analysts. As often as not, a typi- cal batch terminal in a network, rather than being a "dumb" terminal such as IBM's i^ou, will be a mini or micro processor based device providing local tile storage and processing capabilities as well as the capacity to read cards and print paper. he dec with 1 the de er tha ysts . ty of some ce , a ire a but the phone. ision nteli cisio n the If dial iow- n end micr progr iarg Th entra tr ibu be sions 1 sit n whi e wi inati 1 Com to acquire a igence may, in n of the end decision of n there is the ing into the speed teletyp user may dec ocoraputer lor am It to corarau er system vi us, the decis lized star n ted processin ven turtner r made by analy e. As an exam ch a network 11 turn next on ot the Ke puter Network termi- fact , user etwork possi- system e-iike Ide to home nlcate a the ion to etwork g net- emoved sts at pie of might to a ntucky IKEC- IMt K.tNrULK.K hUUCATlUNAL CUMPUTER K to p vice tion Coram the bate capa were late base inte ECNt rovi s to s o onwe orig h a bill not ly d o ract T was establ de academic r the eight p 1 higher edu alth of K.entu inal design c nd interacti ties, intera actually est yt>. The ba n an IBM jY ive system i ished in computer ublic in cation 1 cky. Ai ailed fo ve proc ctive se ablished tch syst U/lbt> an based lyM ser- stitu- n the though r both essing rvices until em is d the on a bi lAbiiibT Newsletter, Vol. ^, No. J (Summer 197«) ON-LINE I ; on-line: STORAGE \ || STORAGE A higure 4: Hierarcnicai or Iree Network UtCSystem 10. The original batch system developed as a centralized star network and had its origins in tne use western and others had already had or the state's system at the bureau of Computer Services in t-ranktort, the state capital. The establishment of the lull network was conceived ol' as taking place in stages or phases, with the tirst phase connecting tne eight institutions into the computers at Louisville and Lexington over a common communications system. The original idea behind the first phase IS illustrated in Figure !? . The second stage was to expand the original communications system to that depicted m higure 0. Ihe network currently approximates Fig- ure b, although variations in design have taken place. In parti- cular, the node labeled BCS (Bureau of Computer Services, hrankfort} is not now linked into the system at all, and may never be. Ihat deci- sion has been a political rather than a technical one. bigure b: Kentucky Educa Computer Network, Fhas tional e 1 To illustrate the way in which the system has developed we might look more closely at the link between Western Kentucky University (WKU) and the system. Western constitutes a communications sub- lASSIST Newsletter, Vol, 2, No. 3 CSumraer 197a) Figure b: Kentucky tducationai Computer Network, h'nase ii center in the system usin time-aivision multiplexor to ai a ytiUU paud line into lour baud channels. There are two centers on Western's campus eac which, until recently, have one of the i?4UU baud channels w go to Louisville and are switched over to Lexington location of the IBM 3Y0/lbb). ray otate University IMU located further to the west linked to tne multiplexor at via a £?4UU baud line into a t channel. The fourth channel turttier time-division multipl into eight JUU baud channels interactive computing at Wester vide HbT h of used hich then (the Mur- au; , is WKU nird IS exed for n . We cont i direc estab Lexm two e upgra Intel IbUU the u Paduc are no guratio t link lished gton, a vents : de its ligent from an se ot ah Comm w sh n as tat bet mov We HBT term IBM the unit ifting just d ybUU ba ween e made stern' s to a inai I. jYttU) vacated y Colle slig escr ud; West nece dec high to a and en ge. htly the ibed. A IS being ern and ssary by ision to er speed Harris to allow annel by tor 1 puting , diverts System dependin sent a " sent tim in dire another , for the of the i computer puting sometime also be tern thus other th Lexingto tucky Un access t terminal central i into a 1 It wo in Bowli microcom use , b automati two larg ing a lo thence state ' s situatio time of technolo enormous nteractive or a modem in the user to eit 1U or the 1 g upon whether 1" or a "2". e the two compu ct communicatio although that future. In ad nstitutions als s for administ and it is pro in the futur hooked into the providing com an those at Lo n. Already N iversity's IBM he system as a Clearly wha zed star networ arge distribute on-line Louis her the BM jYU a use At the ters ar n with is pi dition , o have rative jected e these KECNtT puting uisvill orthern jYO/n high t was o k IS tu d netwo cora- ville UbX- /Ibb, r has pre- e not one anned all local com- that may sys- nodes e and Ken- b can speed nee a rning rk. uld ng U pute e ab call e sy cai bei syst n ha writ gica cos be possible reen to obt r system fo le to prog y access eit stems on K.EC call to the ng linked em. Altho s not devel ing of this lly feasibl t to the end for someone ain a small r personal ram it to her of the NtT by mak- University , into the ugh such a oped at the paper it is e without user . While a considerable amount of time and effort has entered the planning of KtXNb.T changes have been made in the original design. bti iAi>6i5T Newsletter, Vol. d, No. i l^iummer lyyaj 5ome 01 the changes were made necessary as a tunction ot politi- cal aecisions inMuencing tne lund- itig of the system, some changes were made on the basis of the availability ot hardware tcommuni- cations) at unusually good prices, and some were made as a result of user institutions requiring better or ditferent kinds of services. Unce operatio to restr opment . wn 1 c h m industry for pubi mile ot state go example , cost ot vate ent cost dif differen citic a depend in designm system . a networ do with a lar n It ic t t bu ight may ic ag lease vernm is 1 equi V erpr 1 feren t dec pproa g o g a In a kin ity ge IS n he r rthe be not enci d t ent ess alen se . tial isio ches n w pub ny e oper network ot alway ange of rmore , good to be equ es. Th elephone in Kent than one t servic becaus s in pho ns conce might hether lie or vent, on ation, w comes into s possible Its devel- decisions r private ally good e cost per lines tor ucky, for third the es to pri- e of these ne service rning spe- be made one was a private ce we have hat can we NKTWUKK UFPUHTUNITIES There are a number of advantages land some disadvantages; in using networks. We will note three opportunities provided by such sys- tems: tconomy ot scale; decentral- ization ot organization; and shared applications. Kach of these will be considered in turn. equipme numbers the cos vide ea ger , mo sors an ind ivid resembl tems . systems same ar the lat tcon not onl If mul the sa necessa copy of for eac number reduce storage base, a in havi gle dat nt of ts ch u re p d pe ual ing Alt is gume ter across users . It was ser wi ower fu r ipher cost that hough steadi nt can halt o reia by poss th 1 , ce als , s t on the ly de stil f the tively 1 so sprea ible to ccess to ntral pro yet to somet smaller cost ot s cl ining , 1 be mad lyyus. arge ding pro- lar- ces- hold hing sys- mall the e in omies y in tiple me d ry to the h pot 01 CO the , ma nd ot ng mu abase of sea hardwar users atabase mainta data ra ential pies ca cost ot intenan her att Itiple wi th m le are e , howev need ac , then in only ther tha user . be re expens ce ot t r ibutes users o ultiple obtained er , for cess to It is a single n copies When the duced we ive disk he data- inherent 1 a sin- copies. Ihe key cons mg hardware be and communicat more important of communciatio cost of provid on the same s without incurri munications. A these cost dif become so grea end of networ while the argum works based on are still impor may come a time scale argument iderat comes ions is wh ns IS ing lo cale ng the t leas ferent t as t ks . ent in an ec tant a when no Ion ion as comput- less expensive costs become ether the cost exceeding the cai computing as a network cost of com- t in the lyyUs lais have not o require tne Consequently , tavor of net- onomy ot scale nd true , there the economy ot ger has merit. Economy of Scale As has already been noted, one 01 the earliest reasons tor net- works was the ability to spread the costs of large scale computing Decentralization of Organizations If, indeed, economies of scale can be achieved through networking, then as large-scale organizations tend toward decentralization sucn lAbiiibT Newsletter, Vol. d, No. i CSuraraer 197B) econoioies become ever more important. An apparent contempo- rary tact of lite in large-scale organizations today is a tendency toward at least geographical decen- tralization and possibly adminis- trative decentralization. Ket many large organizations continue to require access to common data, pro- gram libraries, and ttie like. As a result, witnin a large organization a computer network can provide tnose centralized services neces- sary, and at the same time help maintain coherence and standardiza- tion within a largely decentralized system. fcach individual component of the organization may be doing local computing, as well as remote computing on a network. Particu- larly in higher education this argument nas merit. At least in Kentucky not one of the eight state institutions of higher education would have, or could individually atford , computers of the size and performance of those m KECNET. Applications bnared Hesources including personnel resources, it should then become possible for remote sites to devote relatively more time and attention to user problems rather than to the prob- lems of maintaining hardware and keeping it up and going. Computerized Conferencing and other Opportunities Unce a network has been establ- ished certain opportunities are provided which were not previously available. Une such opportunity is computerized cont erencing . compu- terized conferencing is a process by which "groups may communicate about complex problems through mem- bers' terminals instead of tele- phones" CMcKendree, 197tt) or in person. because of the computer's capacity for maintaining ongoing proceedings, the conference may extend for days or weeks or longer. There are several advantages wnich might accrue trom such conlerenc- ing: 1 . Participants are treed trom constraints of time. Noted numerous times above is the tact that with networks come shared data, software, and hard- ware. Within the context of a com- puter system there is another very major resource constituted by the Skills ot the people associatied with the system. When working within the context of a network it is no longer necd*ssary for every local unit to provide staff with all possible skills, since each remote location nas the possibility of maKing use of the skills of those at other remote sites or at the central site if one exists. as a consequence ot the possibility ot making full use of all the resources available in the system. ^. inlormation related to the discussion is auto- matically processed and stored . i. Text, statistics, and/or cases are printed in easily used anc' standard- ized formats and are available as needed. 4. Periodic voting on propo- sals raised in the dis- cussion may be accom- plished with speed. b. Conterees may be coupled to various modeling, simulation, or other com- - b/ iAbbibI Newsletter, Vol. ^, No. i (Summer lyya; putationai and planning tions tor communications pro- resources, such as Uelphi cessing. UATAMAllUN ^^ (June Method studies. 1976), IIY-IPO. Kelly, Neil, et al . Data Communi- Computerized conterencing would not cations, Kart d. iNKUSiblt.Mb De possible witnout networks. e^ (March lyrO, i":>-^<^ . inrough the use ol' conferencing it Kelly, Neil, et al . Information is possible to reduce travel time Networks: a management update. and costs yet provide the basis tor INi- USKSlKMii