The Irascible ProfessorSM

Irreverent Commentary on the State of Education in America Today

by Dr. Mark H. Shapiro
Man is still the most extraordinary computer of all...  ...John F. Kennedy (May 21, 1963).
Commentary of the Day - Jan. 22, 2000: Will PeopleSoft® Sink the CSU?:
The California State University system, all 23 campuses plus the Chancellor's Office, is engaged in an ambitious project to integrate all of the campus enterprise computing functions into one seamless system of data management.  It is envisioned that all the campuses would use a common set of software tools to manage payroll, human resources (faculty and staff hiring, benefits, etc.), business functions (purchasing, accounting, etc.), and student services (admissions, grades, enrollment, financial aid, etc.).  At present, the various campuses in the system use a wide variety of software systems and computing platforms to manage these functions.  This makes it difficult for the Chancellor's Office to obtain information from the campuses in a timely manner for assessment and planning purposes.  From an economic standpoint it would seem to make good sense to try to use an integrated set of common software packages to carry out these functions, particularly if the new software would give staff throughout the system access to needed data from any computer connected to the Internet.

The Chancellor's Office has selected a vendor, PeopleSoft, to implement its vision of this vast, seamless software system.  PeopleSoft is a fairly large software company with substantial experience in the area of commercial enterprise systems.  Their payroll and human resources systems are widely used and highly regarded.  However, they have somewhat less experience than some other vendors in some areas such as student services software.

Unfortunately, PeopleSoft has run into major difficulties attempting to provide similar integrated systems for a number of large universities in the midwest.  The major complaints have focussed on delays and cost overruns in the installation and configuration of the software, and on poor performance of some of the packages that have been installed.  The situation has become so frustrating that seven high officials from "Big Ten" universities have written a letter of complaint to PeopleSoft about the situation.  In addition, Cleveland State University, according to a story in The Chronicle of Higher Education, is considering bringing legal action against PeopleSoft to obtain compensation for delays and cost overruns.

The system that PeopleSoft will implement for the California State University system is far more complex than any of the ones that already have run into problems.  This, in the view of the Irascible Professor, does not bode well either for the CSU or for the taxpayers of the state of California.  In the IP's view, the chances are slim that the proposed integrated software system can be brought online without extensive delays and cost overruns that could run into the hundreds of millions of dollars.  The reasons for this have little to do with either the ambitious visions of the Chancellor's Office and PeopleSoft, or with the quality of PeopleSoft's products.

Rather, the problems that will be most difficult to solve arise from a number of historical factors, and from the rather unique set of constraints that university enterprise software must meet.  On the historical side, over the years most university information technology operations have been very thinly staffed compared to their commercial counterparts.  This has led to a situation where most software problems have been solved on an ad hoc basis.  I.e., products and computing platforms have been selected to address an immediate need.  Thus, most universities have a hodge-podge of legacy systems using different operating systems, file standards, and hardware.  Often the systems have been cobbled together with a number of software patches unique to a particular campus.  Little thought has been given to an overall information technology architecture on the campuses.

On the constraint side, it essential that much of the information that is collected by these systems has to be kept readily available for decades.  This is particularly true in the area of student records and personnel records.  Universities also are held to very stringent data privacy requirements.  Thus, the security of the data collected by these systems is a major issue.

The first major obstacle faced by PeopleSoft is the conversion of literally hundreds of millions of data files in different formats and on different platforms to files that can be accessed by the PeopleSoft software with no loss of critical information.  In principle, it should be possible to solve this problem.  However, the resources and time needed to carry out the data conversions can be truly daunting given the wide variety of software and hardware systems now in use throughout the the 23 campus system.  The problem is further complicated by the lack of trained staff who have the time to work with PeopleSoft to help make the needed transitions.  On most campuses the existing staff barely is able to keep up with the ongoing demands of increasing enrollment.  Using outside vendors to assist with the transition has been tried with only limited success at other universities.  The problem seems to be that only the campus information technology staff people have the knowledge needed to help with the migration to the new systems.

The next major obstacle faced by PeopleSoft is the highly irregular demands that the pace of campus life places on some systems.  Software that may work well in a commercial environment where demand is more or less constant from day to day can be brought to its knees by the press of a one or two week enrollment period for tens of thousands of students.  Again, this is a problem that, in principle, is solvable.  Existing campus systems cope more or less adequately with these problems now.  However, the integrated software systems that are being proposes are far more complex, and put correspondingly heavier demands on the hardware.  The present systems are mostly independent, so a heavy demand on the student registration system, for example, does not cause a problem with payroll or human resources data processing.  However, in the integrated environment that the PeopleSoft software uses this may not be the case.  Heavy loads on one part of the system could cause serious processing delays elsewhere.  In addition, the new software includes many more options and features than the older system.  This puts even heavier demands on the hardware.  It is far from obvious that the CSU could afford to purchase adequate computing power for all the campuses to meet their peak needs for data throughput.

The third major obstacle is the maintenance of data security and privacy.  In the present "balkanized" system a relatively small number of staff people have access to critical information.  Security is enhanced further for many of the systems by their isolation from the Internet.  As these systems become more and more integrated, and as data flows from campus computers to central data repositories over public networks, the possibility of security breaches increases.  Again, these are all problems that, in principle, can be solved.  However, the time and effort needed can be very costly.

The system is on the horns of a dilemma.  Growth in enrollment is likely to swamp the existing systems on many campuses within the next few years.  It makes good sense to try for a more integrated solution to the enterprise information technology problems.  However, the prospects for implementing system-wide, fully integrated enterprise software within a reasonable time and within budgetary constraints seems very unlikely.

What are the alternatives?  Perhaps parallel integration of common functions might work.  For example, integrating the payroll systems across all the campuses could be an initial goal.  Then, perhaps implementing a common purchasing and accounting system might be the next goal.  Common human resources and student management systems probably are further down the road.  However, if something less than total integration is acceptable, then it may be possible to choose separate vendors for the different functions.  An advantage of this approach is that the system then could purchase the most effective software package and hardware for the particular task, while still achieving a relatively high - if not complete - level of integration.


The Irascible Professor invites your comments.
©2000 Dr. Mark H. Shapiro - All rights reserved.