Planning for Integration Copyright 1990 CAUSE From _CAUSE/EFFECT_ Volume 13, Number 2, Summer 1990. Permission to copy or disseminate all or part of this material is granted provided that the copies are not made or distributed for commercial advantage, the CAUSE copyright and its dateappear, and notice is given that copying is by permission of CAUSE, the association for managing and using information resources in higher education. To disseminate otherwise, or to republish, requires written permission. For further information, contact CAUSE, 4840 Pearl East Circle, Suite 302E, Boulder, CO 80301, 303-449-4430, e-mail info@CAUSE.colorado.edu PLANNING FOR INTEGRATION by Karin Steinbrenner ************************************************************************ Karin Steinbrenner is Director of Computing Services at Framingham State College. Born and educated in Germany, she moved to the United States in 1968, where she has held positions as programmer-analyst, technical director, director of data processing, and manager of information systems in private and public organizations. ************************************************************************ ABSTRACT: The computing services department at Framingham State College relied on UNIX and a trust in evolving industry standards when it took on the task of creating an integrated computing infrastructure to overcome incompatible hardware and databases. In 1987, hardware and software systems at Framingham State College in Framingham, Massachusetts, were at the end of a dead-end street. Statewide standardization of administrative systems in the early eighties had left Framingham State with an aging mainframe incompatible with other campus computers. Many administrative and academic departments had created their own separate computing islands that were incompatible with each other and central administrative systems. Because of outdated technology, upgrading existing systems to include state-of-the-art development tools and an integrated database was not possible. The computing services department was left with two alternatives: either limp along with the existing systems, or totally revamp the entire computing infrastructure. We chose the latter approach. Within a two-and-a-half-year time frame, we planned for and implemented UNIX-based networked computing systems consisting of a central file server with a single campus-wide database and twelve distributed office automation systems. All new computers support the same standard operating system, user interface, and office automation software. Existing PCs, local area networks, and academic computers are in the process of being connected to a single campus-wide network. This article details the progress and problems we encountered in starting from scratch. Isolated Computing Islands Framingham State College is a public, four- year, liberal arts college with about 3,200 full-time students and an equal number of part- time students. The computing services department is responsible for the planning and coordination of all computing on campus. The director of the department reports to the vice president of administration and finance, and works closely with administrative and academic advisory committees. During an effort to standardize computing hardware in all state institutions in Massachusetts, the state purchased a series of CDC Cyber systems to be installed in public colleges and universities in the early eighties. A CDC Cyber 815 was installed at Framingham State to support local administrative and academic computing. The Cyber supported about forty academic users, including a twenty-five-terminal student lab, and thirty administrators. During the same period, a statewide Higher Education Communications Network (HECN) was created, maintained by the Regents Computing Center (RCN) that is part of the Board of Regents of Public Higher Education in Massachusetts. All public institutions of higher education are linked to HECN, providing access to academic computing resources at the RCN and throughout the state, to the state's financial systems, and to nationwide networks such as BITNET. Using COBOL and CDC's file management utilities, the programming staff at Framingham State had developed online student information, billing, housing, and alumni systems to run on the local Cyber. The RCN created system-wide software for admissions and financial aid on its own central Cyber. Complicated procedures had to be developed by Framingham State's programmers to download and upload data between the two systems and to reconcile any inconsistencies between the local and remote databases. On-line access to student, financial, and personnel information was provided to clerical staff for data entry. A few managers requested and obtained access to the system, using the data for audit and control purposes. The general method for data dissemination, however, was with reports requested through, and controlled by, the registrar's office. Payroll and financial systems, except for student billing, have been maintained by the state's Office of Management and Information Systems (OMIS) on a third system: an IBM 3090 running MVS and TSO. The OMIS systems can be accessed for data entry and retrieval only; no data uploading or downloading from or to local systems is permitted. In 1987, crucial components of our administrative systems were running on three different computers, with one -- the IBM 3090 -- totally incompatible with the CDC systems at the RCN and our campus. Data files could be transferred between the RCN and Framingham State. But payroll, personnel or financial data needed locally were maintained either on paper or had to be reentered into local, isolated databases. With the increasing popularity of microcomputers, administrators and faculty had been purchasing stand-alone micros, creating their own local databases and systems to meet their computing needs. This added to the conglomerate of incompatible systems and data. The limited access to financial, personnel, and payroll data on the state's system resulted in multiple micro-based employee name and address files, micro-based payroll systems for verification and control purposes, and spreadsheet- based subsets of the state-maintained accounting system. Plan of Action A long-range computing plan was initiated by the director of computing services, and important contributions were made by the academic and administrative computing committees, the vice president for administration and finance, the Board of Regents staff, and the computing services department staff. The plan called for a new, integrated computing environment to replace the Cyber and integrate all existing and new computers on campus through a campus-wide network. The plan's goals were to make campus computing more democratic -- provide everyone in the institution with an equal computing opportunity -- and to allow information to flow in any direction, rather than upward only. In realizing these goals, computer use would be extended from its then-current base to all faculty and staff, and eventually to students. With such a broad goal, our inability to save and upgrade existing application software and hardware offered us a rare opportunity. Unlike other organizations that have to add an additional computing layer to integrate a variety of existing, incompatible computers, we could start from ground zero to plan for and build an integrated, homogeneous computing structure based on industry standards. The planned computing infrastructure would include: * a central system to function as a file server for an integrated, institution-wide data and knowledge base; * several multi-user office automation systems for departmental computing, with transparent access to the central system; * a campus-wide network that would link all existing and new computers and create a single computing environment; * a gateway to the HECN network to facilitate access to the RCN, OMIS, and national networks; and * a user interface that would provide all users with a single user interface from which they could select the computing capability of their choice. The user interface would facilitate access to such applications as word processing, spreadsheets, electronic mail and messaging, and institutional information where authorized. The planned computing infrastructure would be multilayered, with individual computing done on personal computers, departmental computing carried out on departmental systems, and institutional computing handled on the central system, all transparent to the user. Through gateways, access to the HECN network and the state's financial systems would be provided. To implement a multilayered computing structure with a single user interface, we needed a single multi-user operating system that was supported on a wide range of hardware from many vendors. UNIX was the only operating system that met these requirements. Integral communication capabilities of UNIX would facilitate electronic mail and messaging, centralized maintenance of remote systems, and access to external networks. While UNIX was not as widely accepted as a potential standard operating system two years ago as it is today, its potential was obvious to professionals who had followed the evolution of the computing and communications industry and the growth of UNIX. At that time the number of vendors supporting UNIX along with their own proprietary operating system was increasing steadily, responding to the spread of multi-vendor, incompatible computing structures. Based on our long-range computing plan, we issued a Request For Proposals (RFP) in the spring of 1988. The RFP demanded industry standards where they were either available or evolving: * The central and departmental systems had to support UNIX V version 3, including the Remote File Sharing and Network File System extensions. * All new computers had to support the same user interface that would provide any user on the network with transparent access to any computing capability on campus. * The OA systems had to utilize the 80386 Intel chip. The UNIX operating system running on the OA systems had to support the MS-DOS option to allow for ease of transition from existing MS-DOS applications to UNIX. * The central file server had to support the database management system selected and supported by the Massachusetts Board of Regents for all institutions of higher education. * The network specification asked for a campus-wide Ethernet network with a fiber-optics backbone supporting the de facto standard networking protocol. * Running in the UNIX environment, WordPerfect was the required word processing software and a LOTUS 1-2-3 compatible software was specified for spreadsheet processing. * To minimize disruptions during the transition to new systems, the specifications called for the Cyber to be connected to the network so that any user who was switched to the new network would continue to have access to existing administrative application software. We consciously decided to provide users with basic character mode terminals, rather than sophisticated graphic workstations. The initial installation would then provide computing access to as many people as possible, rather than provide sophisticated graphics to a few, selected users. Implementation Hardware installation began in November 1988. In May 1989, the network, central system, and six office automation systems were installed. We selected a Multimax 320 from the Encore Computer Corporation as the central file server. It is a symmetrical multiprocessor with eight processors, 48 Mbytes of memory, and 3.6 Gbytes of online storage. Encore supports standard UNIX V version 3.2 that complies with AT&T's System V Interface Definition (SVID). Encore modified the UNIX operating system to take advantage of the file server's parallel architecture. The system was delivered with an Ethernet channel and supports the TCP/IP protocol. Access to the file server was facilitated over the network either through intelligent terminal servers or other computers with Ethernet network connections. We purchased 80386-based micros running Interactive's UNIX V version 3.2 for the office automation systems. The 80386 system proved to be an excellent choice. With 8 Mbytes of memory, it easily supports ten to twelve concurrent users at a cost that cannot be matched by any proprietary multi-user system. A ten-user system including four printers, OA software, an MS-DOS option, an Ethernet connection and communication software, sixteen serial ports, installation, a one-year warranty, and two days of training for all users cost about $20,000. Terminals chosen for the 80386 use standard-serial, multi-port connections. Each OA system purchased had sixteen ports in addition to the standard serial and parallel port, allowing for eighteen connections per system, which can comprise a mix of terminals and printers. Most of the existing MS-DOS software purchased runs on the system using the VP/ix feature of Interactive's 386 UNIX. MS-DOS software using bitmapped graphics is limited to the system console. Several network versions of existing MS-DOS software were tested and will support multiple users running MS-DOS under UNIX on the OA system. We completed implementation of the initial network connecting three buildings, the central file server, the existing Cyber, and six OA systems by the end of May 1989. Since then we've doubled the number of OA systems, extended the network to two more buildings, and connected many additional terminals to the network using intelligent terminal servers. The total number of users currently connected to the network is about 200. Plans include extending the network to the remaining campus buildings and adding fifty more users, mostly faculty. A TCP/IP-x.25 gateway was installed in January 1990 for direct access to the state's x.25 HECN network. The new system allows hardware and software maintenance of remote OA systems to be carried out over the network. Backup of all systems is now performed at night onto a disk on the central system. The New System Until it is removed in July 1990, the Cyber mainframe will continue to be an integral part of the new network. All users connected to the network have access to the existing administrative systems until the systems are replaced by newly developed application software. More end users, therefore, now have more access to administrative systems than before the network was installed. And despite being switched from a direct terminal connection to a network connection, original administrative system users never lost a day of access. We chose Relational Technology's INGRES for a database management and 4GL-development system to create a campus-wide database and new application software. The Massachusetts Board of Regents selected INGRES for the system-wide standard DBMS and 4GL. Using INGRES ensures application software compatibility between Framingham State and other colleges and universities in Massachusetts. The programming staff and user departments together developed a new database dictionary using the data files and items of the existing administrative system as a foundation. They wrote programs that map the existing data files into the new database structure, facilitating complete data transfer from the Cyber to the file server before the implementation of any new application software. As new systems are developed, they are now tested in a live data environment, eliminating the need for test data and providing a realistic test environment. Taking advantage of INGRES's standard query capabilities, QBF, the same information that was available only through application programs on the old system was available for querying on the newly installed central file server by July 1989. As a training project and a vehicle to establish programming standards and naming conventions, the entire programming staff developed a documentation system to keep track of various INGRES objects and complement INGRES's own documentation tools. We now have standards documentation for programming conventions, menus, security implementation, naming conventions, and tool kits. After completing the database design and query systems, our programming staff, with one programmer assigned to each major user department, started building INGRES applications for admissions, registrations, personnel, and student billing, to replicate existing functions. Peripheral systems, such as housing, alumni, police, and inventory, are being developed by student programmers under the supervision of the programming manager. By mutual agreement, each user department dedicated the time of one contact person -- a "partner" -- to share equal responsibility with the programmer for the new application software. Development of the new software evolves through prototyping, initially using QBF to test screen layouts, and then migrating to 4GL code. As the software is developed, ongoing dialogues with the end user ensure that expectations of the user are met by the system. Early on in the development stage, users were given access to the new administrative systems, to allow them to review, test, and provide feedback to the programming staff on an ongoing basis. An added advantage of this approach is that users know how to operate the system when it is placed in production. Evaluation By insisting on standards for hardware, software, and networking, we have been able to implement a lot of computing power for a relatively small cost. The new computing and communications infrastructure can be maintained by a small staff (ten FTE), because all elements of the system are supported centrally via the network, and hardware and software are consistent throughout the system. The cost of the initial system configuration was half that of an equivalent configuration proposed by one proprietary vendor. Through a participative maintenance agreement for the central file server, and an innovative maintenance contract for the OA systems, the yearly maintenance cost of the hardware and software combined is less than 5 percent of the total system cost. Adherence to standards eases the expansion of existing hardware and creates an ongoing opportunity for competitive pricing by keeping our system compatible with a variety of vendors. No overhaul of an entire computing infrastructure can be implemented without problems. The single largest problem we had, and still have, is our small support staff. While we expected a strain on our resources, Murphy's Law often prevailed. Our ten full-time staff people had to train themselves on a new operating system and system software, and immediately become trainers and supporters for 150 computer-illiterate users. Existing production software needed ongoing support while new hardware and software was implemented, all within a two-year time frame. Fortunately, user departments were very cooperative and agreed to freeze the status of existing production systems. Administrative and academic departments and the senior management of the college fully supported the project. Often non-computing services department staffs endured hardship without placing the blame on the computing services staff. Two computer science students assisted with the programming effort, while other students were employed to help with hardware and network installation. Throughout the implementation process, the entire computing staff remained enthusiastic despite some minor setbacks and doubts along the way. The department secretary assisted with user support and report programming, the operator in charge of maintaining and running production systems assisted users who encountered "hung" terminals, programmers converted word processing documents: everyone pitched in where help was most needed. Encore, as the file server vendor and supporter of INGRES, provided excellent UNIX training and support, as well as more than the contractually-agreed level of INGRES training. The vendor of the office automation systems, a single-person start-up company, provided outstanding UNIX and networking support. He often assisted with problems that were not related to the OA systems, and during the height of the implementation, virtually lived on campus and worked side by side with our technical support staff. No attempt was made to design new application software, but rather to maintain the functionality of the existing production system and convert it -- with improvements -- to a system using an underlying database. The learning curve of INGRES was longer than anticipated. While INGRES provides a very powerful application development environment, it takes a long time to learn and to gain expertise. The initial plan to utilize INGRES QBF for simple table maintenance tasks did not work out, because QBF is too limited, even for the simplest of applications. Attempting to replace an entire computing structure and philosophy within two-and-a-half years is an experience that one wants to live through only once. The process was exhausting and put more stress on everyone involved than anticipated. However, now that users and technical staff are seeing some results, the struggle seems to be justified. Without this sink or swim approach, Framingham State would have stayed with outdated and fragmented equipment and software, which would have made any improvements difficult, if not impossible. The initial benefits of the networked computing structure lie mainly in the area of better communication between staff, faculty, and departments. Framingham State will realize much greater gains once an institutional knowledge base is developed and made accessible to all staff levels. The entire institution will share, work with, make decisions from, and explore new concepts and methods from a single, consistent, and up-to-date set of knowledge. This knowledge base will likely be expanded beyond traditional database boundaries to include campus events listings, calendars, policies, and procedures, as well as campus bulletins and newsletters published by various departments and interest groups. With the evolution of better and less expensive hardware and software, Framingham State will enjoy more sophisticated workstations, more powerful OA systems using the 80486 chip -- possibly with multiple processors -- and a graphics user interface. Planned are the migration to INGRES's Intelligent Database that includes extended data management facilities, the ability to manage objects and embed rules directly into the database, and the potential physical distribution of the database using INGRES STAR. For academic computing, an interface is envisioned that will connect a user to any local or remote computing resource through selection from a multi-choice menu. The successful implementation of Framingham State College's long- range computing plan is the result, in large part, of its adoption of industry standards. The same UNIX operating system for all computers facilitates centralized system-level support; networking standards ensure connectability on campus and to remote networks; a single user interface and office automation system reduces the human resource requirements for end-user support; and a relational DBMS supporting industry-standard SQL allows for application software sharing with other public higher education institutions in Massachusetts. Most importantly, relying on standards provided us with the best hardware and software selection at the lowest cost without holding us hostage to a single vendor. ************************************************************************