CAUSE/EFFECT

This article was published in CAUSE/EFFECT journal, Volume 22 Number 4 1999. The copyright is shared by EDUCAUSE and the author. See http://www.educause.edu/copyright for additional copyright information.

Decentralized Information Technology Requires Central Coordination!
by Sarah C. Michalak, Julio C. Facelli, and Clifford J. Drew

This viewpoint addresses the development of information technology (IT) organizational infrastructure in institutions with a high degree of independence at the department and college level and a high degree of local autonomy in the development of information technology resources. What does an institutional history of extraordinarily decentralized computing mean for 21st-century technology support, and how can necessary changes be made in a deeply rooted institutional culture?

Computing environments in many institutions of higher education may be characterized as decentralized to some degree, with greater decentralization in large research universities.1 This tends to be true of decision-making structures also when viewed along a continuum by institutional size and mission. Larger institutions with a significant research mission tend to emphasize disciplinary academic efforts in both instruction and scientific investigation and often place a great deal of authority at the local department level.2 Decentralized computing environments support the academic mission of such institutions strongly. However, when IT services become fundamental to the daily operation of the university, that is, when they become the equivalent of basic utilities, they require centralized coordination to attain maximum reliability and universal equity of access.

Although some electronic resources are increasingly viewed as academic utilities, moving them to central management or coordination represents a substantial change in some universities. In those universities, in many cases the local academic departments initiated the early evolution of computing in almost the same way that they fostered growth in support of their individual disciplines. Consequently there may be a high level of departmental ownership and reluctance to relinquish control despite the fact that further growth and development of these resources may well depend upon support extended by the institution as a whole.

Understanding the history

It is important to remember the roots of our highly decentralized computing environments as we seek improvements in service quality and efficiency. In research universities the basic sciences, engineering, certain medical specialties, and other subspecialty professional areas have a long history of providing computing facilities and services. For these fields the use of computing resources has been part of the daily routine and, to some extent, individuals in these fields have difficulty understanding why the “newcomers” make so much fuss about technology. After the introduction of mini and microcomputers in the late 1970s, development of computing capabilities in these areas was often funded and directed almost entirely from within the departments or colleges where they were used with little consideration for the overall impact on the institutional IT infrastructure. The institutionwide priority to develop centrally provided foundation services remained low in many research universities.

As the use of technology continued to spread to all corners of the campus, development and funding often followed the same model as before, with pockets of effort or islands of capacity emerging within local academic units. However, increased use of information technology in every area of campus life and its use for broad-based communications altered its role within these universities. What was once a novel means of communicating among computer scientists or engineers became fundamental for vital institutional operations as well as for academic exchanges. Electronic resources are now used in parallel with, and in many cases as replacements for, telephone and paper communications. Users expect these services to perform with a utility-like reliability that an entirely decentralized IT infrastructure cannot support. Therefore, provision of reliable IT services must assume as never before a central role in the institutional infrastructure,3 and IT must remain central to the planning, budgeting, and policy development of the chief academic officer, deans, and chairs.

Advantages and disadvantages

While we recognize the significant advantages of distributed computing environments in decentralized academic institutions, we also recognize the disadvantages that can result from secondary effects.

To provide effective IT services, the institution must fulfill the real and perceived needs of individual users to the greatest degree possible. Greater diversity in an institution increases the challenge of meeting individual user needs as the number, type, and customization of services increase. In a distributed environment many of these needs can be addressed by local modifications or by locally modified applications. Localization can be accomplished readily because access to IT professionals is direct and involves a minimum of bureaucratic processes. From the user perspective the problem has been solved because the adverse effects that local modifications may produce in the institutionwide infrastructure do not have an immediate effect on the individual’s work. The secondary effects of local modifications will emerge later in the form of higher costs and/or less reliable systems. The average user will not see a relationship between costs, reliability, and requests for local customization.

The proximity to and accessibility of IT personnel is another important factor in the perceived responsiveness of distributed environments. Because individual disciplinary specialists can make pedagogical and operational decisions that are most appropriate for their group, for most faculty and staff distributed computing appears to be more appropriately responsive to local needs than a centralized computing environment. The need for coordination only appears later when institutionwide crises develop due to the lack of coherence in the system. Personal contact with the departmental “computer person” has in itself a sort of placebo effect that may disguise these larger issues.

Over time very large systems accumulate significant in-house adaptations to meet real or perceived institutional needs. When either the hardware or software used in these systems becomes obsolete, it is difficult to migrate or upgrade the systems in a manner that preserves all of the necessary adaptations. Dropping the local modifications implies a change in the established work processes that is greatly resisted in all organizations but particularly those where decentralized decision making prevails. In distributed computing systems the adaptations are more localized and it appears easier to migrate to new systems because the individual systems are much simpler. However, this argument ignores the secondary effects that these changes may have in other interconnected systems.

Good distributed environments are planned

The advantages of distributed environments often seem to be offset by the problems they entail. We believe that the difficulty is not with the distributed environment itself, but with uneven quality of implementation and with the common misconception that the benefits of natural evolution will in themselves outweigh accompanying difficult technical and organizational problems. Furthermore, viewed from the local perspective, the distributed environment appears to be less complicated than the centralized environment, providing another good excuse for avoiding the extra work of coordination.

The most compelling argument, though, to explain the problems with distributed environments can be found in realizing that good decentralized IT environments are planned; they do not evolve naturally.4 An effective distributed environment is responsive to user needs and is designed from the user’s perspective. Effective IT service requires clear definition of the services to be centrally provided and/or coordinated and those which must be controlled and directed locally. This definition must take place at the highest possible level in the institution and should include mechanisms to enforce accountability of the units providing central services. Clear definition from institutional leadership can also eliminate unnecessary competition between central and local units. Coordination must be pursued constantly and seriously. To encourage coordination, staff resources and central subsidies should be allocated in a strategic fashion during the budget process. When the coordination process is not taken seriously, very uncoordinated and almost chaotic distributed environments evolve. Several circumstances that typically contribute to this type of development are discussed below.

Lack of leadership in academic and administrative computing

In highly decentralized environments there may be no central administrator making institutional decisions or informing administrative leaders about how computing fits into overall university strategies. In these situations the overall IT organizational structure is not framed by the university’s leadership and therefore no institutionwide corporate understanding develops regarding the manner in which academic and administrative computing relate to each other or how computing fits into the academic framework. Where computer-literate departments, such as science and engineering, attempt to establish that framework, other members of the campus community may view them as self-serving or as trespassing on departmental turf. Yet a double bind can result when other nontechnical departments attempt to lead. They can be seen as newcomers without the proper portfolio to challenge long-established IT expertise.

Lack of central funding for system coherence

In the complex organization of research universities, colleges, departments, and research groups obey a strong instinct to solve problems locally. The prevailing reward systems which favor personal excellence in scholarship are inadequate to encourage development of infrastructure. It is the responsibility of the central administration to allocate funding to counter this trend, achieving a healthy balance between centrally offered incentives and local initiative. University administrations which have not become vested in a progressive IT environment may fail to understand the importance of maintaining this balance and as a result may treat any IT-related unit as one of several departments competing for funding. In these circumstances, the IT providers lose out over the academic units. To create a reasonable balance between central control and local initiative, it is imperative that university administrations consider central IT units as complementary and supporting agents of the academic mission, not as their competitors.

Uncertain role of the central computer center

As the age of the mainframe faded, some campus computer centers had a crisis of identity. In institutions in which there was a central vision for the direction of IT, the computer center was reshaped to fit the challenges of the networked environment. In institutions in which that vision was lacking, the computer center was left to compete for funding with all other academic and administrative units. To survive, computer centers required a competitive advantage--a charter (and funding) from campus administration. In institutions without strong computer centers, departmental entrepreneurs stepped in to create pockets of expertise. In some cases, the result was a community culture split between technology “haves” and technology “have nots” (usually undergraduates.)

Poor performance of central services can lead to creation of new units

To compound the problem, where institutionwide computing leadership was lacking, poor performance by some central computer centers was not corrected and “poor services” became an expectation. Some central service units were not held accountable for good customer service, leading to a culture of unresponsiveness. Incorporating accountability for service quality requires a transformational change in the prevailing culture.

It is often viewed as easier to create another unit to perform a given function than it is to resolve a performance problem in an existing unit. Extremely distributed decision making makes it easy to create such units almost at will, and without coordinated oversight of IT functions, many such operations emerge. Locally provided e-mail and networking services may multiply. One of the longer-term ripple effects is the difficulty in achieving coordination or consolidation because territoriality develops rapidly and quickly sinks deep roots. On an institutionwide basis duplicative services are confusing to users (especially students), are excessively expensive, and are unreliable.

Technicians can become de facto IT policy makers

In a competitive market for qualified computer staff, there is a constant tendency to try to “make them happy.”5 This is even more important when the staff resources are highly distributed and most units lack second tier and backup personnel. The loss of “the computer person” in a department can produce serious disruptions in the operation of the unit.6 Under these circumstances, management of academic units tends to be accommodating to the aspirations of individual technical personnel, which in many cases are not aligned with the academic or other operational goals of the department. This problem is exacerbated, in many cases, by the lack of technical understanding on the part of academic management of the unit. Under these circumstances the necessary balance between technical innovation (technicians’ agenda) and services to the users (operational or academic agenda) becomes seriously compromised.

In an environment in which the role of the campus computer center has become marginalized in favor of local development, there might easily develop a belief that all central coordination is at best suspicious and at worst untrustworthy and pernicious. In a competitive academic environment, clinging to this distrustful perspective can become almost an intellectual necessity: for these faculty and staff, surviving and thriving computationally requires them to focus strictly on their local point of view. From this position develops the tendency to leave to the low-level, departmental technician the ultimate responsibility for departmental IT direction.

Critical reasons for centralizing coordination

A research university characterized by these conditions must reorient its IT environment because existing infrastructure cannot support current demands and may collapse under the pressure for advanced and universal services. We have identified the following as the most critical reasons for change.

Increasing expectations that IT will be focused on user needs

IT is increasingly more broadly viewed as academic infrastructure for the total institution rather than as the concern of the science, engineering, and health science sectors.7 An emphasis on solving user problems may have been less crucial in previous computing eras because the major users were assumed to be literate in computing. Less experienced computer users are not familiar with the fragility of most systems and they quickly become dependent on these technologies for critical day-to-day operations, without a full understanding of their limitations. IT management cannot neglect this challenge. We must develop the robust, dependable, and ubiquitous IT infrastructure that our users demand.

Need to maximize use of scarce talent

The shortage of IT professionals will continue for many years. Universities are ill positioned to compete for these workers with higher wages because paying market equivalent salaries will greatly alter the salary hierarchy of academe. Universities will have to compete by using their largest source of talent--students.

This requires higher levels of vertical integration of the IT workforce with less experienced professionals and student assistants working under the supervision and mentorship of experienced professionals. Eventually these professionals will leave the university environment and will be replaced by their apprentices. The required levels of vertical integration of the IT workforce cannot be achieved in an extremely decentralized environment because of the additional competition among individual units within the institution. Note that when the pool of talent is depleted, the “richer units” will also suffer because their most capable people will leave academe without leaving a legacy of trainees. There must be a pipeline to keep the supply of talent coming. Moreover, effective vertical integration can reduce overall personnel and training costs by requiring fewer highly qualified people and encouraging specialization. With good institutional training programs and good planning, vertical task integration for IT staff can be accomplished in an environment of central coordination and local control, particularly if territorial borders remain soft.

Increasing security threats

All of higher education has observed a dramatic increase in dependence on IT resources for teaching, research, and administrative tasks. Daily operations of higher education institutions are increasingly dependent on IT resources, and any significant disruption of these services due to malicious attacks can be devastating to the institution. The security vulnerabilities are multiple including unauthorized access to sensitive information, damage of information, and disruption of services. Security is an inherent central function. There is no such thing as distributed security because the security of a multi-element system is only as strong as its weakest link.8 Moreover, because the university as a whole is the registered legal entity, the institution implicitly assumes all liabilities due to lack of security in the departmental units. It is also important to note that departments are typically less able and/or less willing to commit resources to security.

Increasing regulatory compliance and reporting requirements

In a highly decentralized institution, oversight bodies have difficulty knowing where to direct communications regarding technology-related matters. To outsiders (and sometimes to insiders as well) it is not clear which of the many relevant administrators and computer professionals should be contacted if higher education stakeholders need information about computing needs or performance on campus. Those needing to contact institutional representatives include regents, trustees, state commissioners of higher education, legislative analysts, and many others. Without such information it is difficult for the university to compete properly for funding or to remain compliant with governmental regulations or reporting requirements. Visible central coordination of information policy and enforcement communicates to the internal and external communities that IT will be managed responsibly and for the general good.

Increasing demands for IT services without new funding

Demands for improved or additional IT services become increasingly urgent regardless of whether new resources are available. Static budgets are not unusual in higher education and they have an enormous impact on IT operations.9 Higher education IT funding is limited compared with other industries that have been able to achieve productivity gains through technology. It is vital that we maximize economies of scale in a manner that does not have a negative impact on academic or research programs. IT budgets are perceived as a sinkhole by many administrators. This is not surprising because funding requests appear from every corner of the institution and grow constantly more clamorous. Rarely are measurements taken to indicate how additional funding has enhanced the mission of the institution.10

One of the most promising tools to mitigate budget increases in IT is to introduce economies of scale by standardizing hardware, software, and support services. Many elements of the IT environment have been developed under the “one of a kind” mentality which is problematic because it leads to environments that are extremely costly to support across campus. The efficient choice is to develop “standard solutions” and make them attractive to users in a variety of fashions. Perhaps initially this can be accomplished through central subsidies, but ultimately it must be driven by utility services that are more reliable and of higher quality than those that can be achieved through decentralized operations. In this scenario deviations from the standards that may be necessary for educational or research needs will have to be fully justified.

Recommendations for change

If a university community has not found the right balance between central coordination and local control, the need for change will be obvious, even for many high-level administrators without an intimate knowledge of IT. The challenge for IT administrators is to provide realistic solutions to problems that can be implemented without major disruptions of services and that are fiscally and politically feasible. We have identified a number of principles that should guide the migration process.

Change cannot be mandated

Unless there is considerable a priori agreement across the institution, any solution or effort to promote change from the top will fail. Universities are notorious for resisting mandated change, but fortunately there is also willingness to recognize and follow leadership and excellence. If the migration toward a more centralized and/or coordinated environment is properly presented and explained on its own merits, the community is far more likely to embrace the change.11 A properly planned and conducted information campaign must then be accompanied by excellence in the execution of the plan and through appropriate financial incentives.

Academic administrators must join the effort

Administrators in the academic sector of campus must see and “buy in” to IT as an essential part of the academic infrastructure. Several reasons support this assertion.

First, academic budgeting within most large R1 institutions is characterized by a great deal of distributed decision making. Department chairs and division heads construct their budgets and forward them to their respective deans. College deans typically have a great deal of autonomy to work with those budgets. Budgetary adjustments are made in order to emphasize or de-emphasize activities that the deans believe are in the best interest of their colleges or are most likely to gain favor with the university administration. By the time budgetary recommendations have reached the vice presidential level, they have achieved a considerable level of solidification. If IT is to be a pervasive academic tool, emphasis needs to begin to be reflected in budgets long before they reach central administration, certainly at the dean level and preferably beginning with department chairs. Such acceptance is more prevalent in disciplines with a history of computing infrastructures although it is spreading to others. However, such acceptance has not yet begun to include allocating local budgets to ensure ubiquitous central services.

A second important reason for fostering academic administrator “ownership” relates to the need to support central infrastructure budgeting. For example, academic deans represent a level at which advice and support are often sought by central administration for a range of initiatives. It is important that deans see the enhancement of central IT infrastructure as benefiting their academic efforts as well as those of the broader institution.

Plan strategically to identify effective coordination areas

Reversing many years of decentralization and neglect of the IT infrastructure is not easy. Moreover, significant energy and resources are used to deal with constant crises and in confronting assertions of why change is not possible. This is not an uncommon circumstance in higher education technology circles, and strategic planning often takes a lower priority than daily survival.12 Such planning, however, is vital in order to maximize the effective use of resources. The amount of energy and funds required to make substantial changes on a broad front is well beyond the budgetary resources available to the institution without causing significant dislocations in other programs. Under these conditions change must be gradual. The scarce resources available have to target those areas in which centralization and/or coordination will produce the biggest returns in terms of improving services, gaining political capital, and producing monetary savings that can be reinvested in the process. These decisions should be made with great care and at the highest possible level in the institution.

Encourage policy makers and faculty to control local IT environments

IT decisions are complex. In most cases it is not possible to disconnect policy and technology issues. Under these circumstances it is tempting for “technologically challenged” faculty and administrators to delegate the decision-making authority for IT to low- or mid-level technicians with narrow agendas. It is imperative that senior IT administrators become highly visible and trustworthy to convince faculty and administrators that they must spend significant amounts of their time and energy planning for a more coherent IT environment.

Conclusion

In highly decentralized institutions in which there has been a long, undirected evolution of the IT environment, incoherence of vision and direction may seem natural. Ultimately, however, that incoherence can be destructive. Those who are dedicated to changing the negative aspects of decentralization without sacrificing the valuable characteristics of local initiative and creativity must create change through coaching, mentoring, and strategic thinking. Ultimately, however, there must be a match between leadership at the middle and top levels. Even as they provide moral support for fundamental change, top leaders must be willing to invest new resources and must reinforce the patience and direction of the mid-level leaders. Where there are vision, strength, and resources at both the top and middle levels, an institution can recreate an environment in which instructional and information technology can fully inform the university’s teaching and research missions.

Acknowledgement

This viewpoint grew out of a presentation made at the task force meeting of the Coalition for Networked Information, April 26-27, 1999, Washington, D.C.

Endnotes

1 M. Santosus, “Degrees of Change: The New Demands of Higher Education Are Pushing Up the IT Grading Curve,” CIO, May 1997, 78-86.

2 M. K. McGinn and W. Roth, “Preparing Students for Competent Scientific Practice: Implications of Recent Research in Science and Technology Studies,” Educational Researcher 28, no. 3 (1999), 14-24.

3 K. C. Green, “When Wishes Come True: Colleges and the Convergence of Access, Lifelong Learning, and Technology,” Change 31, no. 2 (1999), 11-15; and President’s Information Technology Advisory Committee, Information Technology Research: Investing in Our Future (Arlington, Va.: National Coordination Office for Computing, 1999).

4 M. Ringle and D. Updegrove, “Is Strategic Planning for Technology an Oxymoron?CAUSE/EFFECT 21, no. 1 (1998), 18-23.

5 P. Freeman and W. Aspray, The Supply of Information Technology Workers in the United States (Washington, DC: Computing Research Association, 1999).

6 S. Jurow and B. L. Hawkins, “IT Staff Shortage: A Crisis that Calls for Collaboration,” CAUSE/EFFECT 21, no. 4 (1998), 11-12; and A. West, “The Information Technology Staff Crisis: Plan for It!CAUSE/EFFECT 21, no. 4 (1998), 13-18.

7 B. L. Hawkins, “Developing the Necessary Infrastructure,” Educom Review 34, no. 3 (1999), 56; and M. Pittinsky, “Studying Today to Envision Tomorrow: The Future of Enterprise Academic Computing Systems,” Educom Review 34, no. 3 (1999), 56.

8 G. Bruce and R. Dempsey, Security in Distributed Computing (New York: Prentice Hall, 1997).

9 Freeman; Hawkins.

10 R. N. Katz, “A Time to Assess?Educom Review 34, no. 2 (1999), 56; and Pittinsky.

11 G. L. Anderson, “Toward Authentic Participation: Deconstructing the Discourses of Participatory Reforms in Education,” American Educational Research Journal 35 (1998), 571-603.

12 Ringle.

Sarah Michalak (michalak@library.utah.edu) is director of Marriott Library, Julio Facelli (facelli@chpc.utah.edu) is director of the Center for High Performance Computing, and Clifford Drew (drew@gse.utah.edu) is associate vice president for instructional technology and outreach in the Office of the Senior Vice President for Academic Affairs at the University of Utah.

...to the table of contents