APPLICATION BRIEF

ADEMIC:ORMATIONSTEMS

Carnegie MellonUniversity

Reachjjjg for World Leadership inEducational Computing ^^Communications

:32s till

/Aw

«

7s?

arnegie Mellon University's historydates to 1900, when industrialistAndrew Carnegie founded Carnegie

Technical Schools in Pittsburgh's Oaklandsection. The schools became a degree-granting college in 1912 and were renamedCarnegie Institute of Technology. A 1967merger with the Mellon Institute ofResearch led to the current name andstructure.

Carnegie Mellon enjoys internationalrecognition for its emphasis oninterdisciplinary learning. This reputationfor excellence covers such areas asengineering, technology, science, liberalarts, fine arts, and public and privatemanagement.

Carnegie Mellon is home to numerousresearch centers exploring a variety offields, among them robotics, biology,computer science, magnetics technology,CAD/CAM, and document design.Significantly, about 85% of CarnegieMellon's 6,000 students regularly usecomputing in their educational activities.

Carnegie Mellon Pursues Its Vision of"The University as an Information Utility'

Carnegie Mellon University, one of the mostcomputer-intensive campuses in the world, is also aleader in the development and deployment ofautomated academic communication systems.

Through its emphasis on communications,particularly on integrated network methods andtools, Carnegie Mellon embodies the academic idealof the university as an information utility—aninstitution which exists primarily to collect, produce,and disseminate information. To achieve this end,Carnegie Mellon's efforts are well advanced toeffectively link all computers on campus into onecommonly shared communications system.

The university also is working with like-mindededucational institutions that are striving to establishstandards for shared computer networks andeducational software on a global scale.

Computing Demand Helps Develop Concept

As early as 1981, computer use on campus had becomeso widespread that the university was finding it difficultto keep up with the demand for computing cycles on itstraditional time-shared mainframe computer.Departmental systems were proliferating and, by 1984,students were buying microcomputers at a rate that,together with the university's own rapid build-up ofworkstation clusters, would soon result in 5,000computers on campus for some 7,000 students, faculty,and staff—all without campus standards and all withlimited interconnectivity.

Carnegie Mellon President Richard M. Cyert respondedin September 1981 by creating a multi-disciplinary taskforce to consider the proper role of computers oncampus. The task force recommended that CarnegieMellon aggressively direct the trend. The group alsostressed the importance of communications amongcomputers as a means of efficiently sharing informationand resources. In looking at the popularity ofmicrocomputers, it further suggested that it was time toexperiment with a major shift to personal computing.

The recommendations of the task force—rec-ommendations bolstered by earlier experiences withnetworking within the Computer Sciences Department—stimulated the administrative decision to create acampus-wide network of all computers and informationsystems.

RICHARD M. CYERT IS PRESIDENTOF CARNEGIE MELLON UNIVER-SITY.

"We were interested in seeing the campus so saturatedwith computers that there would be one per person,"Cyert notes. "We had faith that it would lead into someinteresting developments. We also wanted to takeadvantage of the trend by developing educationalsoftware for students. But the original driving force wascommunications. We always thought in terms of thenetwork, and never were really interested in havingstudents randomly bring in personal computers."

He explains that the decision was made early to providecomputing services as a centrally directed campusutility—making intelligent workstations generallyavailable to everyone. The objective was to have anoptimum synthesis of time-sharing and personalcomputing.

"When we started," Cyert continues, "we realized wewere possibly talking about the world's largest local areanetwork. Even after talking to IBM management aboutjoining us on the project, it wasn't clear that it waspossible. But, after a four-month feasibility study, bothparties decided to go ahead with it."

Cyert emphasizes that the school did not set out todesign a network with specific ideas on how it would beused: "We took a purely technical approach. We saw itas our job to provide an environment—a set of tools thatour students and faculty could use as they pleased. Wealso decided to let them continue with their ownsystems. It thus became the administration's problem totry to adapt the departmental devices and software to thenetwork, rather than the other way around."

(JiriWkM.'lloil

EDUCATIONALCOMPUTING

From Concept to Creation:An Exercise in Ingenuity

"The problems were immense: conceptually, socially,and technically," says William Y. Arms, Vice Presidentfor Academic Services. "It's a university-wide projectwhich interacts with many other universities. We haveover a hundred different makes of computers on campus,innumerable types of network systems, workstations,and so forth. The scale of the project is unprecedented."

The "unprecedented" effort was code-named "Andrew,"after Carnegie Mellon benefactors and namesakesAndrew Carnegie and Andrew Mellon. (Andrew,

AS PART OF ITS GOAL OF ONECOMPUTER FOR EVERY PERSONON CAMPUS, CARNEGIE MELLONHAS CREATED A SERIES OFINFORMATIONAL BROCHURES ONCOMPUTING THAT ARE DIS-TRIBUTED TO ALL INCOMINGSTUDENTS.

WILLIAM Y. ARMS, VICE PRESI-DENT FOR ACADEMIC SERVICES,DESCRIBES CARNEGIE MELLON'sEFFORT ON THE ANDREW PRO-JECT AS 'UNPRECEDENTED.'

developed under a joint development agreement, isowned by IBM.) In October, 1982, Carnegie Mellon andIBM teamed up to create an on-campus softwaredevelopment group called the Information TechnologyCenter (ITC).

ITC's mission was to develop the central part of theAndrew system, the network, file system, andworkstation software. Meanwhile, others at CarnegieMellon were developing the interfaces that would tieCarnegie Mellon's polyglot of microcomputers in variedlibrary, telecommunications, computing, andinformation systems into one neat and transparent digitalcommunications package.

Some 30 people were assigned to the ITC to design anddevelop Andrew—20 Carnegie Mellon employees,including the project director, and 10 IBM employees.

Concurrently, Carnegie Mellon assigned strong supportroles to several existing campus groups. TheComputation Center would deploy and maintain thesystem; the Center for Development of EducationalComputing would support courseware development; andthe Telecommunications Office would install cablingsystems throughout campus. Simultaneously, a socialsciences committee was formed to study the social,organizational, and educational effects of the Andrewproject. To this team, then, went the challenge to havethe network ready for general use by fall 1986.

Four years after it began, the vision has proved to bemore than a dream: "It works," says Arms. "In fact, theeffort is now starting to pay off with some significant,demonstrable successes. But that is not to say our earliervision was flawless."

He points out two big differences between what wasaccomplished and what was planned. "At first, there wasan emphasis on striking a balance between personalcomputers and host computers. But as we found thatmore and more things could be done on themicrocomputers, host computers were downplayed."

Arms is quick to add that large mainframes still havesome role to play. But their role as network controllersand file servers diminished as the capabilities ofmicrocomputers increased.

"The other big change," he says, "is that the plansassumed there would be nothing else on campus exceptAndrew. This model ideal was comparable to buildingan expressway across Wyoming, working with virginterritory, when in fact we were talking about running onethrough something like Chicago. The unavoidableinterrelationship with other computer systems oncampus was not foreseen. The way around the problemis to use standard protocols that make the networkaccessible to everyone, but it is a big task."

The ITC: Innovations In Software Development

James H. Morris, Carnegie Mellon Director of the ITC,agrees that the Andrew project has been influenced inunforeseen ways by the rapid growth of microcomputingpower.

"The Carnegie Mellon program had its earliestbeginnings with something called the SPICE Project,which, among other things, was the Computer ScienceDepartment's specification for an ideal workstation. Weassumed that we would have to take a hand indeveloping such a machine ourselves. Carnegie Mellonconsulted extensively on the design of the IBM RT PC,which exceeds the specification in many ways.

"Basically, the ITC was to produce innovative softwareapplications," Morris notes. "In fact, I would say thatabout half our efforts have supported educationalapplications, word processing, and communications—such things as electronic mail and bulletin boards. Theother, probably somewhat larger half, has gone into thenetwork and the file system.

"Rather than develop a new operating system, we adopted4.2 (Berkeley UNIX*) as our basic operating system fora lot of reasons," Morris explains. "We wanted anoperating system that could run on a variety ofhardware, and was supported by several softwareorganizations."

Morris notes that a massive file system—which is thecentral communications backbone of Andrew—had tobe built underneath 4.2. Files and file servers, scatteredthroughout the network, all work together to provide theillusion of one giant database that everyone on campuscan access.

"The most shocking thing to many people is that thereis no host requirement," says Morris. "Instead, we havedivided the Andrew system into two parts: thebackbone, which we call VICE (Vast, IntegratedCommunications Environment), and the applicationssoftware driving the workstation clusters hanging fromthe backbone, which we call VIRTUE (Virtue IsReached Through UNIX and EMACS)."

VICE, according to Morris, is a collection ofcommunication and computational resources serving asthe information sharing core of a user community. VICE

JAMES H. MORRIS, DIRECTOR OFCARNEGIE MELLON'S INFORMA-TION TECHNOLOGY CENTER,THINKS ANDREW'S EVOLUTIONHAS BEEN INFLUENCED BY THERAPID GROWTH OF MICROCOM-PUTING POWER, PARTICULARLYTHE IBM RT PC, WHICH EXCEEDSCARNEGIE MELLON'S OWN SPECI-FICATIONS FOR A HIGH-FUNCTIONWORKSTATION.

*UNIX is a trademark of AT&T

6

The Communications Network:A Perspective on VICE

DON SMITH, SENIOR IBM ENGI-NEER, LINKED IBM'S TOKEN-RINGNETWORK TO CARNEGIEMELLON'S COMMUNICATIONSSTRATEGY. THE UNIVERSITY ISUNITING VARIOUS LANS AND DIS-PARATE COMPUTING EQUIPMENTACROSS THE CAMPUS INTO ANEFFICIENT SYSTEM THAT CANEXPAND AS NEEDS GROW.

also can be envisioned as one large file system with anumber of services carried out by individual personalcomputer workstations.

"We wanted to avoid tying up the big mainframes withmundane communications tasks," Morris says. "Thissystem does have bottlenecks—which are the computingcapacity of the file servers—but this is easily overcome.What's important is there's no single bottleneck that willbecome white hot as we ramp up this year to 3,000users."

The glue holding VICE together is Andrew's networkprotocols. Because many departmental mainframesalready use DARPA (Defense Advanced ResearchProjects Agency) Internet Standard protocols forcommunications, and because the Berkeley 4.2 UNIXoperating system has implementations of the Internetprotocols, DARPA Internet became the obvious choiceas a standard for Andrew.

Thus, any machine and software configuration can hangfrom the network, but provisions have to be made toimplement those protocols with the 4.2 operating systemvia Internet. In order to communicate with various IBMsystems, for example, the ITC implemented SNA(Systems Network Architecture) under UNIX. Thispackage is used to drive the network's IBM 3820 laserprinters and other shared devices.

Don Smith, a senior IBM engineer who has worjced onAndrew since its inception, was impressed with theeclectic nature of Carnegie Mellon's computingenvironment. "As the Andrew project really got rollingin 1983," he recalls, "there were about 15 Ethernet* andtwo Pronett LANs on campus.

"These were mostly independent networks, spread allover campus in individual buildings. Some were small,but others were quite large. The Computer Science net,for example, had 300 to 400 computers online. And,anywhere in this jumble of networks could be productsfrom a dozen different vendors. Missing was aconnection among the different networks. Also, we hadto have a way to gracefully add dozens of personalworkstation clusters," says Smith.

Hence, from an ITC perspective, four basic pieces of thenetwork needed to be pulled together. First came theneed to define and develop an intelligent workstationapproximating the SPICE specifications. Second was theneed for software standards that would permit

* Ethernet is a trademark of Xerox CorporationtPronet is a trademark of Proteon Corporation

JOHN LEONG, CARNEGIE MELLONASSOCIATE DIRECTOR OFNETWORK DEVELOPMENT,INCORPORATED ELECTRONICDEVICES IN ANDREW THAT ROUTEDIGITAL TRAFFIC BETWEEN FILESERVERS AND EACH OF THENETWORKED LANS.

communications between the various LANs withoutrestricting the continuance or addition of otherwiseincompatible departmental systems. Third, thedistributed file servers and file server disks had to beincorporated to unify them and make them transparent tothe entire Andrew installation. Fourth, a large build-upof microcomputer workstation clusters, utilizing IBM'sToken-Ring Network technology and serving thousandsof microcomputers, had to be introduced into the systemin a non-disruptive manner.

The IBM Token-Ring Network has been operating atCarnegie Mellon since January, 1986. According to JohnLeong, Associate Director of Network Development, theIBM Token-Ring Network has proven to be very stable,and its extensive, well thought-out network managementfunctions have greatly facilitated operation.

An important element of the networking activity is theinterconection of a large number of local area networksto allow widely dispersed workstations to communicateeffectively with the file servers. This is achieved byrouters developed jointly by the Computer Science andElectrical and Computer Engineering departments. The

routers were deployed under the direction of Leong."The routers are base on off-the-shelf hardwarecomponents such as 68000 multibus systems and theIBM PC/AT," Leong says. "They are equipped withappropriate local area network interfaces. The basic ideawas that we could put a router up for interconnectingthree networks at a cost of about $4,000. They are quitecost effective and can handle roughly 500 packets persecond."

According to Leong, "routers are best described as verysimple dedicated processors that look at the headers ofthe data being transfered and then shove them down theappropriate cable, depending on the destination address.Because of the simplicity of the hardware and software,it is a very stable piece of equipment. We have sufferedfew failures over the past two years."

The IBM Token-Ring Network may be a relativenewcomer to Andrew, but it has become the system'sfastest-growing segment. From six installations at mid-year, supporting about 50 IBM PCs and 150 IBM RTPCs, the network now carries some 20 IBM Token-Ringclusters supporting about 1,000 IBM microcomputers.Another 1,000 microcomputers and workstations ofvarious types are supported by approximately 23Ethernet clusters.

MIKE WEST, IBM MANAGER, HASWORKED TO MAKE THE VICE FILESYSTEM COST-EFFECTIVE, EFFI-CIENT, AND EASY TO USE.

VICE File System: An Overview

Mike West, IBM Manager of the VICE file system, hasworked on the project since December 1982. "Our goalswere many," he explains. "First, we wanted it to berelatively inexpensive. We now serve more than 50workstations per server, so that goal was met.

"We also wanted a graceful degradation under load. Infact, we've now gotten VICE to where it slows downand gives us a chance to correct problems, but doesn'tcrash. Our next goal was to achieve performance levelsthat would exceed those of timesharing. This meant thatwe wanted VICE to appear as though all files were local.Next, we established minimum targets for the system of5,000 workstations and 10,000 users.

"Finally," West continues, "we wanted VICE to be easyto operate. We didn't want it only to be accessible tospecialists or hackers. In other words, the system isdesigned to more or less run itself.

"The local file emulation feature was achieved by aprocess of full file transfer," West explains. "Wheneveryou ask for a file in VICE, the entire file is copied toyour local disk. From then on you run with it locally.When finished, you copy it back to the file server. Thisgives us simplicity and high performance. It runs as fastas the local disk, because it is the local disk. It's alsosecure, because a copy of the master file is alwaysretained in VICE—until you update it."

The full file transfer process is combined with asophisticated workstation caching technique to buildAndrew performance. This caching procedure eliminatesthe need for access to the network for small, commonlyused files. They're always resident on the workstation.

"Allowing a user to have secure access to his own fileswas also a major goal," continues West. "Since we donot have control over the workstations, we do not trustthe workstations to authenticate the users for VICE.Instead, during log-on the user is validated to VICE andthe workstation is only trusted as an agent for the user."

To implement much of this, authentication was built intothe remote procedure-call mechanism.

THIS WORKSTATION, ONE OF SEV-ERAL IBM RT PCs CLUSTERED ONAN IBM TOKEN RING NETWORK,HAS IMMEDIATE AND 'TRANS-PARENT' ACCESS TO THEANDREW SYSTEM.

A Perspective on VIRTUE

VIRTUE is the component of Andrew that brings thepower of UNIX and of advanced personal computerworkstations, such as the IBM RT PC, within the graspof the network user.

According to Andrew Palay, Carnegie Mellon Managerof User Interface Software Development, "VIRTUEprovides a consistent user interface across all threegeneral classes of academic user applications: generalutilities—such as text and drawing editors, mail, andbulletin boards; courseware authoring tools—such asCMU Tutor; and specific courseware itself—such as TheGreat American History Machine, which providesgraphic representation of demographic information, andthe Notecard System, a writing development tool."

ANDREW PALAY IS MANAGEROF USER INTERFACE SOFTWAREDEVELOPMENT.

10

The PC Server

JOHN H. HOWARD, IBM INFORMA-TION TECHNOLOGY CENTER SITEPROJECT MANAGER, SAYS THEPC SERVER (RIGHT) GIVES PCsUSING DOS A DIRECT LINK TOANDREW'S UNIX ENVIRONMENT.THE PC SERVER WAS DEVISED BYIBM SYSTEM DESIGNER LARRY K,RAPER.

For machines like the IBM RT PC, which are designedto run the Berkeley UNIX operating systems,connection to Andrew was a relatively simple chore. ButCarnegie Mellon has over 3,000 IBM PCs and othermicrocomputers owned by students, faculty, and staffthat never were intended to run UNIX.

"To give those machines access to Andrew," says JohnH. Howard, IBM ITC Site Project Manager, "we tookan IBM RT PC that connects to VICE as a regularworkstation, and wrote a piece of code that allows theIBM PC to connect in through the IBM RT PC. So youcan go from an IBM PC to the IBM RT PC, and throughthis to files anywhere in the file tree. The PC Serverapplication, according to Howard, accounts for half ofthe IBM Token-Ring Network connections.

PC Server Block Diagram

^ ^ ^ SNAP Conversation

oooo SNAP Control Channel

11

The Andrew Cabling System

Robert Cape, Carnegie Mellon Director ofTelecommunications, has overseen a massive wiringproject that not only links every building on campus, buthas forged extensive links to the outside world. "Whenwe're done," he claims, "we will have the world'slargest local area network, with 7,000 nodes—a nodebeing defined as an intelligent workstation or greater.

"As a ballpark estimate, we're talking about 50buildings, 11,000 outlets, and 300 miles of IBM Type 2Media Specification cabling. If it isn't the largestinstallation of the IBM Cabling System, it's certainly themost significant retrofit to a set of existing buildings."

Selecting a cabling system was no trivial task. Cape'soffice first began grappling with the question of whatwiring strategy to employ some two years ago.

"The turning point came when we found ways to runEthernet on the twisted pairs of the IBM CablingSystem. This gave us the confidence we needed,knowing that IBM's system had the flexibility needed forour complex installation."

The IBM Cabling System at Carnegie Mellon provides,within a single sheath, two pairs of shielded wires fordata, and four pairs of unshielded wire for voice. Thissingle cable substitutes for the multiple coaxial,twinaxial, and telephone wiring installationstraditionally used for institutional communicationsnetworking.

"Concurrently with the IBM cabling, we installedextensive inter-building fiber-optic cabling to completethe connectivity of the campus," Cape adds. "Our mainoff-campus, fiber-optic links run to the Mellon Institute.We also have fiber into the Bell of Pennsylvania centraloffice in Oakland, and use it to extend 64 KB service toworkstations in neighboring homes and apartments overtheir existing loops. This service is part of a projectcalled the Metropolitan Campus Network."

ROBERT CAPE, CARNEGIEMELLON DIRECTOR OF TELECOM-MUNICATIONS, HAS OVERSEEN AMASSIVE WIRING PROJECT THAT,WHEN COMPLETED, WILL BE THEWORLD'S LARGEST LOCAL AREANETWORK.

The IBM cabling project posed quite a differentchallenge of going back to existing buildings and tryingto make a $7-million capital improvement withoutanyone noticing. "Another of the complexities wefaced," says Cape, "is a varied assortment of buildings.They're of all different kinds. The classic view of cabletrays running down a hall, with drops coming off eachside, just couldn't work here. The architecture wouldn'tsupport it."

One alternative proved particularly effective—thedistributed riser, whereby cabling was fanned out in thebasement or the attic, and then distributed vertically—

12

Campus freeport i

People

Names Depts Titles

nwnuMnrleU

1.nntar1!

Workstations

11

inter rants

sl orations

. ' . • • : • : • .

wets

1

1

Map Items

Plan

[ y rn:

Athletics I

Names

Acaclemic3

Parking Bridges

THE ANDREW CABLING SYSTEMCONSISTS OF A FIBER-OPTICBACKBONE (THICK VERTICALGREY BAR) AND SEVERALATTACHED LANS AND FILE SERV-ERS. THIS SCHEMATIC, SHOWN ONAN IBM RT PC SCREEN PRINT OUT,WAS PREPARED BY THOMASPETERS, IBM SENIOR PROGRAM-MER (BELOW).

with wiremold raceways coming off into rooms. This hasbeen particularly effective in residence halls, where noone wanted conduits running horizontally through thebedrooms.

"I had thought that the hardest problem would be to findwiring closets, which in reality are small rooms," heexplains. "I had visions of going to deans anddepartment chairmen asking which junior faculty officesthey were going to give up. But it has turned out not tobe a problem, though we have had to be very creativewith janitors' closets, landings on stairwells, and othercrazy places."

13

Communicating Through Andrew

JONATHAN ROSENBERG, CAR-NEGIE MELLON MANAGER OF THEMESSAGE SYSTEMS GROUP, ISRESPONSIBLE FOR DEVELOPINGAND DELIVERING BULLETINBOARD, ELECTRONIC MAIL, ANDDATABASE SERVICES TO THEANDREW NETWORK.

Now that the Carnegie Mellon Andrew network is inplace and working, how is it being used?

The answers are many, but can be broken down into thethree general areas: messaging, teaching, and research.The first of these responsibilities resides with JohathanRosenberg, Carnegie Mellon Manager of the MessageSystems Group, who is responsible for developing anddelivering bulletin board, electronic mail, and databaseservices to the Andrew network.

"What makes my job fun," Rosenberg says, "is thisgiant time-sharing file system that looks like one bigserver. We have about 12 file servers either online orsoon to be online, with about 120 gigabytes of storage.So, essentially, we have unlimited storage. But thenicest thing is that we can all share this commonsystem.

"The problem you usually run into with electronic mailand bulletin boards is trying to deal with physicallyseparate machines. Getting timely information tomachines and updating them is difficult, because youhave to go into the file system of each to post to thebulletin board, for example. With Andrew, they all share

the same virtual file and have immediate access toinformation or messages when they're posted. Withoutthe system, we would have to worry about updates andsynchronization and locking—but Andrew does it all for

us.

At Carnegie Mellon, as at most universities, Rosenbergpoints out, many departments have their own mailsystems, which they insist on retaining because theyserve unique informational purposes. Rosenbergresolved the issue by leaving the existing messagenetworks intact. He concentrated instead on making iteasy for them to incorporate Andrew services into theirown systems. Individual groups may now have twodistinct hierarchies of messaging service; one whichmeets their specific internal needs, and one whichconnects them to the broader community.

"The definition of a message here is interesting,"Rosenberg continues. "In general, it constitutes ASCII-based (American Standard Code for InformationInterchange) mail. But, in fact, a message might includepictures, graphs, or even spoken text. We have recentlyreceived an award from the National Science Foundationto study the editing and transmission of multi-mediadocuments, and we are looking at those needs."

14

Typescript

Mailbo

torn

freeport} Kal

freeport(freeport) % cd Homevork(freeport) % bush(freeport)%A

Console ViceAndMore [4.8]

QR Dl D2

3:57 PM

Fetching History( 3:55:27 PM)Fetching 28886 1449420 I 3:56:26 PMFetching 2883600262304 3:57-03 PMFetching 28886000 13702 3:57:06 PMFetching 287862355 1501 3:57:20 PM

freeportFriday 10/17/86

Bush Homework freeport

o/cmu/itc/ torn /HoIT ewock i l l story

| SORT |DETAIL|

Drama French Painting

English

B i.bli o E? say:? Re 3 e 3 r ch

CLASS MEMBER;

] suppose Pat eels are inunorsl, caust QityMenoccompleltly jnotd. but tfiey CBr be

iiiifiiifiiiffl;i;H-iit^iBii-iKn-ivmii

freeport

1986

January5 ra t •!! t f 5

1 2 3 45 6 7 S 3 10 11

12 13 14 15 16 17 1813 20 21 22 23 24 2526 27 28 23 30 31

May

1 2 34 5 6 7 3 3 10

11 12 13 14 IS 1b 1718 19 20 21 22 23 2425 26 27 23 23 30 31

September5 n\ t i? t f S

1 2 3 4 5 67 8 9 111 11 12 13

14 15 16 17 IS 19 2021 22 23 24 25 26 2723 29 30

FebruaryS r« t '» t f S

:2 3 4 5 6 7 S3 10 11 12 13 14 IS

16 17 18 13 20 21 2223 24 25 26 27 28

June

1 2 3 4 5 6 78 3 10 11 12 13 14

15 16 1? 18 13 20 2122 23 24 25 26 27 2829 30

October5 n\ t w I t S

1 2 3 45 6 7 8 9 10 11

12 13 14 15 16 |T7]1319 20 21 22 23 24 2526 27 28 23 30 31

March5 [IL fe ',',' t t 5

12 3 4 S 6 7 S9 10 11 12 13 14 IS

16 17 IS 13 20 21 2223 24 25 26 27 28 2930 31

July

1 2 3 4 56 7 8 9 10 11 12

13 14 15 16 17 IS 1920 21 22 23 24 2S 2627 28 23 30 31

Novembers tn t w t t :•

12 3 4 5 6 7 89 10 11 12 13 14 15

16 17 18 19 20 21 2223 24 25 26 27 28 233 0

April5 in t w t f 5

1 2 3 4 56 7 8 9 10 11 12

13 14 15 16 17 18 1320 21 22 23 24 25 2627 2S 23 30

August

1 23 4 5 6 7 8 3

ID 11 12 13 14 15 1617 18 13 20 21 22 2324 25 26 27 28 23 3031

DecemberS m t w t t S

1 2 3 4 5 67 3 9 10 11 12 13

14 15 16 17 18 13 2021 22 23 24 25 26 2728 23 30 31

ReadMail freeportALL MAIL (2391)

xx/ torn

, / ckk•/ tornv ' torn«/ torn

\/ torn

For EDUCOM, Iuse it to see- wheAccordingly, couli

Ettueom & The Campus Program

Thu,16 Oct8613:S4:15edtgahm Thu, 16 Oct S6 10:48:46 edtpromotion of Zip to CMU-side Wed, 15 Oct 86 20:25:53 edt

Wed, 15 Oct So 20:16:57 edt

people Database Wed, 15 Oct So 20:01:05 edt

ks for people database

Undelete

Print

Tin to the Campus program such that visitors canseated, as well as what exhibits are of interest,li whomever has such information?

TCP ; Search

WINDOWING AND MULTI-TASKINGCAPABILITIES OF THE IBM RT PCALLOW THIS CMU USER TO HISCALENDAR, ELECTRONIC MAIL,AND VARIOUS HELP AND INFOR-MATION SCREENS SIMULTA-NEOUSLY, AND THEN—WITH AFEW KEYSTROKES—ACCESSMATERIAL STORED IN A REMOTEFILE SERVER (BELOW). ANDREWCAN DIVIDE A SCREEN INTO ASMANY AS 16 WINDOWS. THISGIVES STUDENTS AN OPPOR-TUNITY TO HAVE A TERM PAPER INONE WINDOW, NOTE CARDS ANDOTHER MATERIALS IN OTHER WIN-DOWS. ANDREW'S WINDOWS AREUNUSUAL BECAUSE THEY DO NOTOVERLAP AND ARE CONSISTENTIN WHERE THEY WILL APPEAR ONTHE SCREEN.

15

Data security is another issue, says Rosenberg. "To helpresolve the problem, a three-phase authenticationtechnique has been developed to ensure that encryptedmessages go to the right people once the packets havebeen handed from VICE to VIRTUE. The techniqueinvolves a combination of passwords and machine-to-machine double handshakes."

Rosenberg believes that bulletin boards are not trivial,fulfilling a vital role as a "giant information utility" thatshould be generally available. The Carnegie Mellonnetwork currently provides access to some 500 bulletinboards, many of which are received via ARPANET.Andrew subscribers can also create and sustain theirown bulletin boards.

"One of the nice things about our system," Rosenbergconcludes, "is that it's hierarchical. When a newbulletin board is generated, subscribers to older bulletinboards are notified of it."

Teaching with Andrew

The primary expertise in courseware development atCarnegie Mellon resides with the Center for the Designof Educational Computing (CDEC). CDEC and ITCstaffs work closely to support faculty members whodevelop courseware.

"The ultimate producers of educational software must befaculty members," says ITC Director James Morris."We and CDEC diffuse technology and tools to them.Our job is to make it exciting enough and easy enoughthat there are no impediments to coursewaredevelopment."

Courseware developed at Carnegie Mellon is publishedby CDEC in a 264-page catalog which lists dozens ofcourse offerings in architecture, art, design, music, civilengineering, electrical engineering, engineering andpublic policy, mechanical engineering, English, history,modern languages, philosophy, psychology, socialscience, biology, chemistry, mathematics, physics, andcomputer science.

Much of the advance in Carnegie Mellon-developedcourseware has been made possible by a project calledCMU Tutor. Developed by Bruce and Judith Sherwood,CMU Tutor is an authoring language that allows non-expert programmers to write educational software on

16

FIVE VOLUMES OF THE ANDREWUSER'S GUIDE HELP STUDENTSAND FACULTY TAKE ADVANTAGEOF THE SYSTEM'S CAPABILITIES.FEEDBACK PROVIDED BY ASOCIAL SCIENCE DEPARTMENTSTUDY IS BEING USED TO MAKETHE USER'S GUIDE A MODEL DOC-UMENT OF ITS TYPE.

Andrew workstations. It has been adapted to Andrewfrom the MicroTutor language developed at theUniversity of Illinois.

CMU Tutor gives full access to the powerful bit-mappedworkstation graphics capabilities of Andrew, includinglabeled graphs, rotatable displays, pattern-filledpolygons and disks, and animation of icons. Rescalingof programs to window size is significantly simplified,and pop-up menus are easily created and managed.Standard calculational structures are provided, and ithandles input from both keyboard and mouse. CMUTutor also provides an extensive on-line referencemanual.

Regular classes in using CMU Tutor on IBM RT PCworkstations are held for Carnegie Mellon faculty andstudents, as well as for faculty and researchers fromother universities.

IN THIS IBM RT PC LABORATORY,FACULTY AND STUDENTS ATTENDCLASSES IN USING CMU TUTOR,AN AUTHORING LANGUAGE THATALLOWS NON-EXPERT PROGRAM-MERS TO WRITE EDUCATIONALSOFTWARE. BRUCE A. SHER-WOOD, ASSOCIATE DIRECTOR,CDEC, AND PROFESSOR, DEPART-MENT OF PHYSICS (INSERT, TOP),DEVELOPED CMU TUTOR WITHJUDITH SHERWOOD (FAR LEFT).SHERWOOD GIVES UNIVERSITYOF PENNSYLVANIA FACULTY MEM-BER ELIZABETH PALAZZI SOMEONE-ON-ONE HELP.

Also driving the revolution are developments like theintelligent tutor that John Anderson of CarnegieMellon's Psychology Department built for teachingLISP. He believes he has evidence that it takes only halfthe time to reach the same level of proficiency in LISPas it takes in traditional lecture classes. He has anothertutor for teaching geometry to high school students inlocal schools. The preliminary evidence is that it raisedgrade averages by one full grade.

17

Other Implications of Andrew

The Committee on Social Science Research inComputing (CSSRC) is headed by Sara Kiesler and LeeSproull, Professors in the Department of Social andDecision Sciences. Sponsored in part by IBM, thecommittee was formed to study the social aspects ofcomputing, as well as the social, organizational, andeducational effects of Andrew. "We are trying to assessthe impact of creating electronic communities," saysKiesler.

"We have two goals. One is to do research onfundamental processes. By this, I mean researchindependent of the particular application you might beusing. For example, how do people communicate usinga computer network? We are looking beyond surfacedifferences to try and find out what's really important increating social effects. Is it the linking up of manypeople, or the reduced social context that matters?"

The other goal is to provide feedback for systems andapplications under development, Kiesler explains. Thedesigners of these systems can use what has beenlearned from studying people at Carnegie Mellon andelsewhere. Considerable feedback has already beenforthcoming in setting up the mail system, and inhelping to devise policies for managing the developmentand implementation of projects.

"We looked at how people really use a system. We haveinvestigated electronic systems used in classrooms," shesays. "We've done empirical research of all kinds, fromexperimental research on how groups use computer mailfor decision making to studies of business andgovernment organizations."

Facts gained from this research help guide systemmanagers and designers in their modifications oradditions to the Andrew network. "The distributednetwork will allow new kinds of work and groupinteractions," Kiesler adds. "Therefore, understandingthese potentials can influence both technical and policychoices."

SARA KIESLER, PROFESSOR INTHE DEPARTMENT OF SOCIALAND DECISION SCIENCES, HEADSA COMMITTEE STUDYING THESOCIAL, ORGANIZATIONAL, ANDEDUCATIONAL EFFECTS OFANDREW.

18

For The Future

Although Carnegie Mellon University has built whatmay be the world's most extensive local areacommunications network, further enhancements arealready being planned.

"On a practical level," says President Cyert, "we stillhave a long way to go in terms of pulling all computingresources on campus into one commonly shared,compatible system. So our development work willcontinue."

To help achieve this, the university is building a strongAndrew Systems Group totally separate from the ITC fordeveloping, installing, and maintaining systemcomponents as Andrew evolves from a developmentsystem into a production system.

Continues Cyert, "On the academic side, we willincrease our efforts to implement the university-as-information-utility idea. We also will push harder forinter-university cooperation on standardizing systemsand software for academia. We must get to the point

where educational software is as portable and usable astext books. Work done at any university should bereadily available to all of academia."

Progress in this latter arena, according to Cyert, is beingmade through the Inter-University Consortium forEducational Computing (ICEC), which represents some20 universities and colleges dedicated to developinghigh-quality educational applications software forworkstation environments. ICEC is considering aproposal that will take the best software from theAndrew Project and projects at other universities tocreate a standard system for academic computing.

"The subject of standardization will also certainly betaken up by EDUCOM, which holds membership inICEC," says Cyert. "We look to IBM and other vendorsto support these emerging standards in their futureproducts, so that the broader academic community canbenefit from our efforts and we in turn from theirs. Wethink the choice of Pittsburgh as the 1986 EDUCOMConference site reflects our growing leadership in thiswork," Cyert concludes, "How better could we serveacademia?"

For further information contact:James Morris, DirectorInformation Technology CenterCarnegie Mellon UniversitySchenley ParkPittsburgh PA 15213Telephone: 412 268-6700

19

ACISIBM Academic Information Systems

International Business Machines CorporationAcademic Information Systems472 Wheelers Farms RoadMilford CT 06460

This publication is intended to dem-onstrate the utility of IBM products,and is not an endorsement of userprograms, third party programs orsystems design. References in thispublication to IBM products, pro-grams or services do not imply thatIBM intends to make these availablein all countries in which IBM oper-ates.

10/86 Printed in USA GK21-0036-00

GK21-0O36-00