PENN PRINTOUT
The University of Pennsylvania's Online Computing Magazine

PENN PRINTOUT March 1993 - Volume 9:5

[Printout | Contents | Search ]


The Wharton School's distributed staff model for computing support

By Michael Eleey

Wherever we turn today, information technology is changing. Hardware costs are declining, while capabilities are increasing geometrically. Operating systems and application software are becoming more powerful, yet more complex. Purchasers are demanding "open systems," the ability to choose the best components from different sources and combine them in a "plug-and-play" environment unconstrained by proprietary vendor standards. And the rate of each of these changes is accelerating.

On campus, these technology shifts and the individual control over information tasks they make possible are being recognized by students, faculty, staff, and administrators. Individuals are turning to the new, cost-effective desktop technology for many tasks, including ones heretofore performed on large terminal-host systems or carried out manually. These trends place great strain on our traditional systems for training and supporting users of computing, and on the systems needed to interconnect the growing variety of devices and activities in computer centers, instructional and research labs, and faculty and staff offices and homes.

In this volatile environment, managing technology means managing change, and we must search for new approaches. Here's a case study of how one Penn institution--the Wharton School--responded by introducing a "distributed" computing support organization. To the extent that Wharton's environment resembles others on campus, its experience may point to some useful directions for redeploying support resources.


The old approach

In 1986-87, the Wharton School began to reexamine its computing facilities and services. Historically, Wharton's approach to academic computing placed responsibility for a common set of general tools with its central computing organization, and responsibility for specialized or local needs not covered in the common set rested with departments and offices. Wharton's central computing group, for example, charged with providing host services for the School's departments and administrative offices, was also assigned responsibility for School-wide microcomputer labs and general microcomputer user services, including training, consulting, and documentation. Departments and offices were responsible for the purchase and maintenance of microcomputer hardware and software for their faculty and staff and to meet particular program requirements. Underlying this approach was the principle that departments, if given the direct responsibility, are best equipped to assess and prioritize their unique needs, whereas a central organization is better suited to address the more widely shared requirements.

Intensive interviews with faculty and administrators in the School's academic departments suggested that this principle was not well implemented: An unacceptable degree of separation had developed between the central organization and the departments. Faculty complained that inadequate computing support hampered their research productivity. They spent too much time tending to the details of their PC hardware and software, couldn't get answers to some kinds of technical questions, and wanted specialized advice and consultation on technology. Department chairpersons and office administrators wanted better management of their hardware and software inventory, needed advice for planning and acquisition of technology, felt that the central computing group was insufficiently responsive, and had difficulty in recruiting, supervising, and retaining quality computing staff within their units. The central computing organization itself found that it didn't have a sufficient understanding of the different departments' needs, plans, and directions, nor did faculty and administrators realize the full range of computing facilities and services that were available to them. Furthermore, successive expansions in host system capacity and support services in 1986, 1987, and 1988 had been quickly consumed by latent user demand.


Searching for alternatives

In its decentralized structure, Wharton is a microcosm of the broader University. Responsibility and authority for budgeting and administration are located at departmental and administrative division levels, and the heads of these units have much discretion in allocating resources to fulfill their unit's plans, so long as these are consistent with School-wide directions. In response to its computing problems, Wharton decided that, rather than abandon the original principle of decentralized management in the case of its computing activities, it would reorganize its computing efforts to apply the principle more effectively. And since traditional methods for planning system capacity and support services were no longer effective, Wharton ceased trying to establish a "final" hardware/software/network service plan. It began instead to focus on building an adaptive solution comprising systems and organizations adept at managing change--mechanisms that would facilitate modification of any part of the technology base at low incremental cost and minimal service disruption.

In its search for organizational alternatives, Wharton considered three common models for managing computing support services:

  • Centralized computing center. Serves a wide population of users and typically provides mainframe-type facilities as well as training, documentation, product evaluation, and consulting for its faculty, student, and administrative users.
  • Information center. Provides training, product evaluation, and consultation to end users for a specific set of packages, typically microcomputer oriented, with emphasis on accessing data from the corporate mainframe. The information center concept was originally a response to, and in many cases an agent of, growth in end-user personal computing.
  • Departmental computing. With the advent of minicomputers, computing facilities and support personnel began to be housed within the end-user department, for the exclusive use of members of that department. A wide range of services focusing on the department's needs are provided. Several Wharton departments had, formally or informally, pursued this approach.

Computing staff roles in these three organizational models are roughly similar, except that somewhat greater variability in staff quality and turnover tends to occur in the departmental model, because of decentralized hiring, restricted intraorganizational career mobility, and supervisory diseconomies associated with the smaller scale of these operations.


The new approach

The Wharton planners recognized that departmental computing, already well established in several units within Wharton, was of continuing importance because of the great variety of faculty computing needs across departments. They also saw from the problem analysis and interviews that improved mechanisms for ongoing communication between the central computing support group and departmental faculty would be essential. They therefore settled on a hybrid organizational structure combining elements of centralized computing and departmental computing. The key innovation was the notion of "distributed support"--placing computing personnel in the departments, but under central management. The plan also called for retaining a central support operation for the core group of technologies and services; to do otherwise would merely shift the burden for these to the departments at greater overall cost, and limit the ability of the distributed staff to provide innovative support.

The heart of the distributed support approach was the establishment of an effective working partnership between the central computing organization and the participating department, with the latter in the role of a valued client or customer. In this partnership formula

  • The central computing group and the participating department jointly develop specific computing support staff job descriptions and select candidates.
  • The client department provides funding for the distributed staff position, with some level of cost-sharing by the central group in the ideal case.
  • Distributed personnel reside in the client department, with space, office facilities, and computing equipment provided by the department.
  • The client department sets overall service priorities, while the distributed staff manage the day-to-day allocation of effort and resources under the central computing group's general direction and technical supervision.
  • Distributed staff attend regularly scheduled meetings held by the central computing group, which serve essential management and coordination functions and facilitate sharing of information and solutions.
  • The department and the central computing group share training and staff development, or negotiate these ad hoc in special cases.
  • The two parties conduct a joint program and staff review 90 to 120 days after the initial staff placement; thereafter, the central computing group conducts personnel evaluations annually, with departmental input. The program is renewed on a fiscal year basis, with departments required to commit to each subsequent year three months in advance, for staff planning purposes.


The pilot program

This distributed support program began with a pilot project in one academic department. The pilot was designed to develop an understanding of the impact of the approach on the organizations involved; to determine the costs, time, and effort required to install, maintain and propagate the solution; and to project the likelihood of success in achieving the objectives and solving the initial problem.

The pilot project demonstrated the effectiveness of the concept in trial use, and in the process revealed how the "ramp-up" period can be shortened if the individual already is familiar with either the central or the departmental organization. In the pilot effort a broad search was undertaken to fill the new distributed staff position, but the selected candidate, preferred by the department's faculty, turned out to be an experienced consultant from the central computing organization. Being familiar with the internal operations and procedures of the central computing group, this individual was immediately able to leverage central services. In addition, being knowledgeable about many of the research computing tools used in the department, the individual was quickly able to develop more in-depth support for the department.

On the basis of the positive results of the pilot experience, the School decided to proceed cautiously with the plan. Central to the implementation strategy was the Dean's requirement to develop the program gradually, using only redeployed resources. The Dean mandated that such partnerships must be initiated by departments, not the central computing organization; there thus was no overt announcement or marketing of the program. Furthermore, departmental participation had to be voluntary; the program (and the staffing expense) would not be imposed on the departments. In addition, departments had to fund the program out of existing resources; no increase in departmental funding was provided by the School for the program. And finally, departments had to provide the position opening from their available head count, since no increase in staff positions would be granted for the program.


The results

The distributed staff program has marketed itself through its own successes, building demand within and across departments. Users and departments have told each other about the program, with the result that departments have initiated contacts and inquiries without direct solicitation.

Participation by academic departments has grown at a steady pace, with one or more departments joining the program each year. Currently the 6 largest of Wharton's 10 departments, accounting for 89 percent of the standing faculty, participate in the program, with a total of seven distributed staff among them. Four administrative units also participate, with five distributed staff positions involved. And to date, no participating department has decided not to renew.

While the specific mix of functions performed by the distributed staff varies by department, the set covers a wide range of support services, including local area network administration, end-user desktop computing support, hardware and software troubleshooting, product evaluation and selection, departmental technology planning and acquisition, technology training, management of high-performance computing facilities for research, configuration of technology in new buildings, support of locally developed applications, and other activities required by departments.


Why it works

Wharton's distributed staff program has succeeded because it fits the School's needs on many levels. First, the approach is tailored to responsibility center budgeting and is consistent with the School's management philosophy and organizational culture. Second, the program facilitates change: It is flexible, adaptive to changing priorities, and provides a framework for continuing innovation and technology transfer. Recently, for example, as the School's major departments began a major transition from mainframe and host-based research computing to RISC/Unix platforms and local area network servers, the distributed staff structure already in place provided a mechanism through which the migration could be effectively deployed, managed, and supported. Third, the approach complements, not duplicates, services provided by the School's central computing group and other computing services provided elsewhere in the University. Fourth, the program provides effective channels for improved interdepartmental communication, coordination, and cooperation, and thus exploits the services and resources of the School's central computing organization more effectively. Finally, the program stays relevant by continuing to evolve, meeting service needs determined by local users over time, as it must do if it is to continue to be judged cost-effective by departmental decision makers.

MICHAEL ELEEY is Associate Vice Provost in the Office of Information Systems and Computing, where one of his major areas of responsibility is ISC computing support services. Formerly he was Director of Academic Technology Services at the Wharton School, where his group implemented the plan described here. He is interested in hearing from readers about other approaches to computing support that are working well at Penn (eleey@crc.upenn.edu, 898-5304).