Framing the System Replacement Decision: A Library Illustration
E. Burton Swanson
and
Carlos Zozaya-Gorostiza
UCLA Anderson School of Management 110 Westwood Plaza | Los Angeles, CA 90095
May 2, 2000
INFORMATION SYSTEMS | WORKING PAPER 2-00 Information Systems Working Paper Series
The IS Working Paper series is a publication of the Information Systems Research Program (ISRP). It provides for the early dissemination of research by the IS faculty, students, and visitors, usually prior to its more formal publication elsewhere. The IS Reprint series includes these more formal publications, many of which supersede the original working papers. To obtain a downloadable index to both series, please visit our website at:
http://www.anderson.ucla.edu/documents/areas/fac/isrp/wp-index.pdf
Comments and feedback on our working papers are welcome and should be directed to the authors. Because their authors typically revise working papers within months of their issuance, we maintain only the most recent three years of the series for distribution. Most of these are downloadable from our website (see above). For copies of older papers, please contact the authors directly.
Publication of the IS Working Paper Series is made possible in part through the generous support of the ISRP by the IS Associates. For further information on the IS Associates, please visit their website at:
http://www.anderson.ucla.edu/x574.xml
Framing the System Replacement Decision: A Library Illustration
E. Burton Swanson1 and Carlos Zozaya-Gorostiza2
1The Anderson School at UCLA 110 Westwood Plaza, Los Angeles, CA. 90095, USA [email protected]
2Instituto Tecnológico Autónomo de México Río Hondo # 1, México D.F. 01000, México [email protected]
Revised May 2, 2000
Copyright 2000 by the authors
Abstract. How should a firm arrive at the decision to retire and replace one of its information systems? Here we offer a proposed method for “framing” the prospective replacement decision by putting it in appropriate environmental context, that is, in the context of the larger industry where application innovations are presently underway. An applications assessment grid is presented to guide the suggested process. An illustrative example of its use in deciding upon the replacement of an existing library management system in a University is described. The proposed method brings together both internal and external logic in assessing whether an application system may have reached the end of its useful life.
1 Introduction
Research suggests that the average life of an information system (IS) within a firm may presently be about twelve years, and that expected life may further be declining (Swanson and Beath 1989; Joseph and Swanson 1998). One study (Swanson 1995) reports that managers among some 54 firms were planning to retire and replace about one third of the application systems in their portfolios within the next three years. Extrapolating from this sample to the broader population, a collective replacement effort of extraordinary magnitude is underway. Yet research on system replacement is on the whole scant and relatively little is known about it and how it actually takes place, or for that matter, how it should take place1.
1 There have been several worthy research efforts. Barua and Mukhopadhyay (1989) and Gode, Barua, and Mukhopadhyay (1990) may be the first to theoretically address the economics of software replacement. Sakthival (1994) and Chan, Chung, and Ho (1994; 1996) offer more recent contributions. Except for Sakthival's broader framework, these works provide normative models focused rather narrowly on when software should be recoded to facilitate ongoing maintenance. They do not address obsolescence or functional enhancement. None are empirically based.
1 How should the firm or other organization arrive at the decision to retire and replace one of its systems? In practice, a system may simply be viewed to be failing and to have reached the end of its useful life. Both users and IS staff may be frustrated with it for any of several reasons. Most basically, when systems become aged their integrity tends to erode and they become more error- prone and problematic to maintain (Lientz and Swanson 1980; Hanna 1993). They may therefore need to be “rewritten” in new code. At the same time, older systems may also require improved functionality and usability. They may further need to be better integrated with other systems in the portfolio. In some instances, they may need to be redeveloped to support revamped business processes. Within the firm, there may thus be an “internal” logic for a system's retirement and replacement.
But systems may also need to be replaced for reasons that originate externally. However comfortable they are to their users, a firm's systems may come to compare poorly to new alternatives arising through innovation in the broader business environment. Certain competitor firms may now have or be developing systems which are substantially more advanced in their functionality and which ultimately deliver more value to the business customer. These firms may be seen as gaining a possible competitive advantage. Elsewhere in the industry, commercial application packages may be purchasable which provide (or promise) superior functionality, usability and maintainability to the many firms now adopting them. Here the industry may be seen as converging around a common business logic for the application (Brady, Tierney and Williams 1992; Joseph and Swanson 1998) with widespread ramifications. In this broader context of the firm's competitive environment, there may thus also be an “external” logic for any application system's retirement and replacement.
Whether internal or external or both, where it is not comprehended well beforehand, the logic for a system's retirement and replacement may eventually force itself upon the firm. For instance, it may become all too painfully obvious that a system is failing to meet urgent user needs or that a competitor possesses a more advanced system which is taking market share from the firm. In such circumstances, perhaps much too often, management may eventually come to act out of perceived dire necessity, playing a game of catch-up which may or may not be successful. Here system replacement can become an act of desperation. “(We replace our systems) when they have screamingly reached end-of-life,” one manager has said (Swanson and Beath 1989, p. 194).
Alternatively, management may plan purposefully for system retirement and replacement. It may institute an assessment process that better anticipates the need. In this spirit, we offer in the present paper a proposed method for “framing” the prospective system replacement decision in terms of its internal and external logic. Central to the suggested method is an applications assessment grid. While further research is required to confirm the method's validity and usefulness, our hope is that it may ultimately enable managers to make their system replacement decisions more wisely and with less desperation.
The paper proceeds as follows. In the next section, we describe the requisite planning environment for undertaking the application assessment. The assessment method itself is next introduced and discussed in its basic details. Following, an example that illustrates the application of the method to the decision of replacing a library management system is described. A concluding section summarizes, offers several caveats and points the way to further needed research.
2 2 The Requisite Planning Environment
The application assessment process suggested here presumes that the firm has an established IS planning environment. For present purposes, we assume that this includes a management steering committee or the equivalent (Nolan 1982) which undertakes an annual review of an IS plan under which the development, implementation, and maintenance of each system within an applications portfolio is articulated and approved over say a three-year horizon. Thus, in this context, each application system already in production is planned for continued maintenance and enhancement over this horizon, or it is planned to be retired and (usually but not always) replaced by a new system.
As a first step in the annual process, we assume that the present established plans for the application systems in the portfolio are given a cursory preliminary review, individually and as a whole. From this preliminary review, a selected few of the systems are subjected to the assessment process described next. Included would be those systems for which the present plans are believed to be most problematic. Among these would be those systems currently planned for continued maintenance but believed to be approaching the end of their useful lives. The assessment process would then aim to “frame” the issue as to whether each selected system should indeed be retired and replaced. It would do so by systematically calling upon both internal and external logic in support of the decision. We describe the assessment process next in it suggested details.
3 The Application Assessment Process
In the context of the planning environment described above, we outline a suggested five-step process for assessing the individual application system preparatory to deciding whether to retire and replace it. The process as a whole is guided by the Applications Assessment Grid shown in Figure 1. As will be seen, the special purpose of this grid is to “frame” the prospective replacement decision by putting it in appropriate environmental and temporal context, that is, in the context of the larger industry where application innovations are presently underway.
Focus Operational Developmental Leading Edge State-of-the-art Innovative Frontier Local Position Current System Future System Global Norming Common Practice Business Convergence
Figure 1: Applications Assessment Grid
From Figure 1, the AAG is seen to have two focal dimensions: environmental and temporal. Along the environmental dimension, the firm's local position with its application is of obvious primary interest. This position is situated both internally, within the enterprise, and externally, in terms of the broader industry. It is situated internally in terms of its acceptance in everyday firm practice. It is put in external competitive context when it is juxtaposed with the industry's leading edge application(s) of the same type, where new innovative ground is presently being broken, and further with “global norming” within the industry, where the bulk of competitors are busy
3 adopting and implementing already recognized “best practices.” Overall, the environmental dimension thus enables the local position to be contextualized in terms of application innovation by both leaders and followers in the industry. (See Appendix 1 for a brief elaboration on the innovation theoretical underpinnings of the AAG.)
Along the temporal dimension the focus is on both operational applications already in production and on developmental applications (including replacements) targeted for implementation within the planning horizon. In the case of the local position, where there is no plan to replace the current system within this time frame, the future system is simply the current system as it is planned to be maintained and enhanced.
The assessment process described next is organized in its steps along the environmental dimension. The process is intended to be applied to each system selected for assessment. Given the purposes of the present paper, we will assume that the system selected is not currently planned for retirement and replacement within the next three years, but that it has been chosen for assessment because current plans have been put into question.
Step One. Preliminary assessment of local position
The suggested process begins with a preliminary assessment of the “local position” occupied by the selected application system. The initial focus is internal and on the system as it is employed and valued among participants within the firm. The system is assessed both in terms of its operational characteristics (the current system) and its presently planned further development over the next three years (the future system). Table 1 provides a check-list of questions which may be asked in the preliminary assessment. We consider first those questions which may be asked of the current system and from what information these questions may be addressed.
Table 1: Preliminary Assessment of Local Position. A Check-list of Questions Current System Future System (Assuming Maintenance and Enhancement of Current System) How well does the system meet the expressed How will the system better meet the expressed needs of its users? needs of its users? How reliably and with what integrity does the How will the system execute its information system execute its information processing processing functions with better reliability and functions? integrity? How well does the system contribute to the How will the system better contribute to the firm's competence in its business processes? firm's competence in its business processes? How does the system create value for the How will the system create better value for the business customer? business customer? What resources are allocated to the system's What resources will be allocated to the system's operation, use, and maintenance? To what operation, use, and maintenance? How will extent are they efficiently allocated? they be efficiently allocated?
From Table 1, perhaps the most obvious question is the extent to which the current system meets the expressed needs of its users. In answering this question, surveys of “user satisfaction” may be helpful (see, e.g., (Baroudi and Orlikowski 1988)). So too should be the recent history of service
4 requests for maintenance and enhancement of the system, as well as the records of any “help desk” which assists users when they encounter problems.
This same documentation should also assist in assessing the extent to which the current system executes its information processing functions reliably and with integrity. “Trouble reports” and downtime reports from operations, and records of corrective system maintenance should similarly speak to this question. So too should any audit reports directed specifically to this issue.
The extent to which the current system contributes to the firm's competence in its business processes can be assessed by ascertaining how the system and its users function effectively together or not in accomplishing basic tasks. Whether users often have to work around the system, rather than with it, to accomplish their tasks is one issue here. Whether users are well trained and knowledgeable in their direct interactions with the system is another. Usability studies are often good sources of information on these issues. So too may be first-hand studies of the task itself, as well as records of “critical incidents” which have arisen in task performance and how these have been successfully handled or not.
Beyond serving its immediate users, the extent to which the current system creates value for the business customer should also be assessed. Such value may be reflected in the low cost of firm products or services therefore favorably priced, or it may be in information services that provide added value, as in the support of after-sales service. Analyses of how the customer would likely be affected in the absence of the current system may be particularly helpful in this regard.
Finally, the preliminary assessment should address the costs associated with the current application system, in terms of the human and machine resources allocated to its operation, use and maintenance. Here IS cost records for operation and maintenance are likely to be available, while records for the cost of system use are typically not. Because the human resources allocated to use may be high, they should be at least estimated. A sample of users may be asked to estimate the amounts of their time spent in direct interaction with the system, for instance, and this may be extrapolated to the whole user population.
From Table 1 it is seen that questions regarding the future system are derived directly from the questions about the current system just discussed. The spirit of these questions is one of continuous improvement, under our present assumption that the current system is planned for maintenance and enhancement over the planning horizon. Accordingly, the answers to the questions should provide the “best case” against which to compare a replacement alternative.
Step Two. Assessment of industry's leading edge
The suggested assessment process next takes an external turn. It involves scanning across the firm's industry and identifying among the competitors for comparative purposes the leading edge application systems of the same functional type. By “leading edge” we mean here those systems that are the most advanced in their functionality and support of the business (often involving an innovative application of new IT). As the term itself suggests, these applications seek to break new ground within the industry and thereby offer their firms a potential competitive advantage. Such undertakings tend to be risky by their nature, as the popular substitute term “bleeding edge” vividly suggests.
5 As with the firm's local position, the industry's leading edge will have both operational and developmental components. The competitors and systems identified in each case may well be different.
Operationally, the leading-edge application system(s) will represent the industry's present “state- of-the-art.” Here the competitive advantage achieved or not by early-adopting competitor firms may be relatively discernable. It should at least be widely discussed among industry observers. Developmentally, the leading-edge application system(s) yet to be delivered represent the industry's “innovative frontier,” when the competitive risks are presently being engaged among the true innovators. These systems may or may not break sharply with today's state-of-the-art.
Because Step Two requires the firm to scan its competitive environment, rather than merely do an in-house assessment, the firm may wish to engage the assistance of consultants. Because of their multiple engagements among firms, consultants are often well situated to assess the industry's state-of-the-art, in particular. The innovative frontier may be rather more difficult for them or others to discern, as it may involve some relatively secret undertakings. Still, news of such undertakings often gets around by word of-mouth. Thus, by paying close attention to current reports and rumors and following up on them to ascertain their accuracy, the firm should be able to gain competitive intelligence on IT applications in its own industry.
Step Three. Assessment of global norming
The suggested assessment process continues with further environmental scanning. Now, however, our attention shifts from the present industry innovators and early adopters of leading-edge applications, to the broad bulk of competitors and their adoption of application innovations which are further along in their diffusion across the industry. We examine not applications leadership, but rather “followership,” reflective of a process which we term “global norming,” an equally important phenomenon in the industrial context.
Global norming in application systems is often relatively easy to observe, because it tends to exert its pressures naturally upon the firm. Again, from Figure 1, there are two forms. Operationally, “common (good) practice” reflects what has already been achieved by the “early majority” of competitors and is now being engaged by the “late majority.”2 Developmentally, the process of “business convergence” represents where the early majority has now turned its adoptive attention. It consists of common practice not achieved, but in the making.
“Everybody's doing it” is the common refrain among the early majority. In this context, “What does everyone else know that we don't? Can we afford to be left behind?” is the yet-to-be- committed management’s worry. The overall effect is that of the bandwagon (Abrahamson and Rosenkopt 1993).3 And, where the application takes the form of a commercial package such as SAP's R/3, for notable example, the process of global norming may yield steeply rising sales for the vendor (Xenakis 1996).4
2 The terms “early majority” and “late majority” are taken from Rogers (1983) and refer to the bulk of innovation adopters who follow the lead of the relatively few innovators and early adopters. See the Appendix to this paper. 3 The bandwagon effect is illustrated nicely by what has been referred to as Airline Magazine Syndrome (Ramiller 1996). The CEO reads about some popular new IT while on an air flight. He or she returns to the office and calls in the CIO. “Why don't we have something like this?” is the question thrown to the CIO. 4 It is interesting to observe that SAP's strategy for the development of R/3 is convergent with the global norming phenomenon. Specifically, R/3 is built on industries' “best practices” as identified by its R&D department
6 Overall, the real importance of global norming is that it may be reflective of a broad change in which the industry as a whole does its business. The adoption of ATM transaction processing by the retail banking industry provides perhaps the classical example. Ultimately, there may be no “competitive advantage” achieved by any of the firms that have adopted the innovation (the “excess profit” or “surplus” may go to the consumer). Yet the logic for adoption is wholly compelling after the fact. The innovation now presents itself as a competitive necessity. New entrants to the industry see little choice but to adopt it. Accordingly, the applications innovations being swept along by global norming within an industry are perhaps even more important to grasp in the assessment process than those which are presently at the leading edge.
Again, consultants may be helpful to the firm's management in the assessment of global norming. Indeed, consultants themselves are important and it must be remembered not unbiased components of the phenomenon (Swanson and Ramiller 1997). Something like Johnny Appleseed, they carry the innovation concept from firm to firm (Dimaggio and Powell 1983, p. 152). Consider the consulting investment in SAP's R/3, for good example (Lieber 1995).
Step Four. Reassessment of local position
With the industry's innovative frame thus established in Steps Two and Three, the assessment process now begins its own convergence. There is first a reassessment of the local position. Both the current and the presently planned future system are reconsidered in the light of the newly- revealed environmental realities and their implied opportunities and constraints. Understandably, the current system may or may not be seen as so acceptable when seen in this new light. Nor may the plans for its future in the form of continued maintenance and enhancement look to be so well conceived.5
Should the present plan for the future system be simply an incrementally enhanced version of the current system, the alternative opportunity to replace the current system with a more locally innovative and functionally superior system, reflective if not imitative of broader innovative trends within the industry, should now be seriously put forth and considered. It is only at this point in the assessment process that the relative costs and benefits of the principal alternatives should be tallied up and put on the table for discussion and comparison.
Should the outcome of the reassessment be a recommendation that the future system now be a replacement system, the proposal for this replacement should be developed in its details and submitted to the last step of the assessment, where portfolio considerations and priorities enter in.
(Bancroft, Seip and Sprengel 1998, p. 35). SAP thus thrives to the extent it makes “best practice” a “common practice.” 5 On the other hand, reassessment may also confirm that maintenance and enhancement of the current system is desirable and should continue. The current system and its plans for the next three years may stand up rather well to the examination suggested here. Note that this puts maintenance and enhancement on a much surer footing than is presently the case in many firms. in the context of the replacement alternative, maintenance and enhancement may now have its own compelling logic. We view this as an important by-product of framing the system replacement decision as proposed here.
7 Step Five. Reestablishment of local position
The last step of the process takes place only when all local positions have been reassessed among the applications. Now proposals for replacement systems must be considered and weighed according to available or securable resources. They must also be considered in the light of their joint effects in delivering value to the firm. Here it is likely that a prioritized development slate would be submitted to the management steering committee for its deliberation and approval. Subject to the steering committee's request for further analysis, the system assessment process is thereby concluded.
4 Illustrative Case: Replacement of ITAM’s Library System
In this section we illustrate the proposed method by applying it to a case in which ITAM, a private university located in Mexico City, decided to replace its library management system. As will be discussed, this decision was reached and implemented very recently. While it was not actually guided by the framework proposed in this paper, it offers us the opportunity to revisit it for illustrative purposes. Its data and the memories of participants were fresh. It involved a detailed assessment that lent itself to being framed in terms of what were the leading-edge and the common practice solutions for Integrated Online Library Systems (IOLS) in Mexico, the US and Europe.
As will be seen, the case involves replacing one package by another in an area where custom code is very rarely used and where global norming is taking place. It thus also corroborates other findings of substantial innovation convergence in business application software (Joseph and Swanson 1998).
Before proceeding, we note that the case focuses only on a single system and its replacement. It does not address the replacement issue in the context of a broader application portfolio, nor does it consider annual reviews within an established planning environment as suggested above. As a consequence, illustrative simplicity is achieved, but at the cost of weakening somewhat the motivation for the assessment process.
Additionally, the replacement issues is seen to be rather clear-cut. ITAM’s current library management system has no future in the organization. Hence the assessment serves primarily to illuminate the broader innovation context in which a new package will be chosen.
Background
The Instituto Tecnológico Autónomo de México (ITAM) was founded in 1946 by a group of businessmen led by Don Raúl Baillères. ITAM’s mission was to contribute to the progress and prosperity of society by excelling in educating people and doing research. During the first years of its operation, ITAM offered only an undergraduate degree in Economics. Later, the Institute opened new programs both at the undergraduate and graduate levels. Currently, ITAM offers eleven undergraduate programs, five masters degrees and one Ph.D. program. As of 1999, ITAM had approximately 250 full time faculty members organized into twelve academic departments, as well as 4,200 undergraduate students and 800 graduate students.
8 The “Raúl Baillères” library has been a part of ITAM since 1946. The library is organized into two units, one located in each of the two campuses of the University. The total physical collection of bibliographic materials is composed of over 71,000 titles, 141,000 volumes and 1,290 periodicals. In addition the library has access to several electronic sources of materials, including Dialog, JSTOR, ProQuest, Lexis, Dow Jones interactive and ACM Digital Library.
The automation of the library started in the late 1980s when several students enrolled in the B.Sc. in Computer Engineering program of ITAM were doing their undergraduate theses. Luis Ernesto Espinoza (1988) developed a system for managing periodicals using the Dataflex database on a personal computer. The system provided some functionality for helping the library staff with various back office tasks but did not include an online catalog for being used by the end user. Due to a lack of proper maintenance, this system gradually became obsolete. In 1994 the library decided to stop using it and returned to a manual operation. A couple of years later, Carlos Gamboa and Rubén Itzkowich (1990) developed a system for managing the acquisitions of materials using dBASE IV on a local area network. However, this system was only in operation for a few months because the users were never adequately trained.
In 1992 ITAM received from the Mexican Council of Science and Technology (CONACYT) a free copy of the MICROISIS system for cataloguing. With this program, the library captured the bibliographic records of theses, working papers, audio and video materials among others. It also bought the LOGICAT package for classifying other materials and for printing the cards of the catalog. By this time, however, having an integrated online library system had become an imperative need. 6
In 1993 ITAM bought Ameritech’s DYNIX system with the justification that this package was widely being used in other libraries around the world. Even though at that time some of these institutions were moving to solutions using newer technologies, ITAM chose to buy a mature package that promised to support the operation of its library in an integrated way and met the needs of medium to large size libraries.
Eventually, several factors contributed to a breakdown of this promise. A first cause of problems was the lack of experience both from the distributor and from ITAM in the implementation of the package. ITAM was the first library to use DYNIX in Mexico and it became the first installation of the distributor with a real customer. The requirements for the implementation were also incomplete and lacked an appropriate level of detail. Further, the legal contract that was signed with the local distributor of the package left ITAM in a weak negotiating position. Many issues related to the installation, implementation and testing of the modules could not be enforced once the project had started.
At the same time, the relationship between ITAM and the local distributor turned problematic. ITAM felt that the technical support personnel of the distributor did not have the level of understanding about the architecture and behavior of the system that was needed. The time required to make adjustments or customizations to the package was always longer than expected. The distributor argued that these changes were not included in the original contract and asked to be paid for any additional modifications to the original code. ITAM felt that the distributor lacked a proactive attitude in solving the problems encountered. The distributor lost credibility with the leading users of the library and this made it even more difficult to reach a solution between the two parties.
6 Appendix II provides a general description of IOLS and their core modules and architecture.
9 A second set of problems was originated when Ameritech delegated the responsibility for maintaining a Spanish version of the software to the local distributor. This version was never fully implemented, and ITAM ended up using a hybrid of the English and the modified version. To further complicate the situation, during the course of the project Ameritech broke its relationship with the local distributor. Since Ameritech was only interested in further supporting the English version of the package, those institutions that had bought the Spanish version from the distributor were now isolated.
Finally, a third set of issues was a consequence of the technology associated with the system itself. DYNIX was one of the first commercial IOLS available in the market. The version bought by ITAM had an internal data structure that was accessible only its developer and distributors. ITAM did not have open access to the data and this situation was aggravated because the technical documentation and users manuals that were provided by the distributor did not correspond to the version that was being sold in Mexico. In most cases, documents were incomplete or badly referenced.
As a result, the system was never fully implemented at ITAM. Only the three most important modules (cataloguing, OPAC and circulation) went into production with only 70% of their functionality available. The other three modules (acquisitions, serials and community resources) could never be used. Thus, at the beginning of the assessment process, DYNIX was being used only at approximately 40% of its true potential.
System Replacement Decision
By early 1998, it was already evident that the supplier of the package was unable to solve the problems encountered. Therefore, ITAM started to evaluate other IOLS that could replace DYNIX. A multi-disciplinary group of people led by Catalina Jaime, the new head of ITAM’s library, was assigned the responsibility of assessing which other package to purchase. Ms. Jaime, whose previous job had been as a librarian for McKinsey & Company in Mexico, was well aware of the limitations imposed by the current system and had a clear idea of what to expect from a new package.
The assessment process took approximately one year. A list of possible commercial packages was initially developed from information included in directories of library automation software (Cibbarelli and Cibbarelli 1998) and comments from other librarians. Detailed questionnaires were sent to the suppliers of leading edge and common practice packages that seemed to better fit the requirements of ITAM. Based on the responses to these questionnaires, a reduced list of finalist packages was identified.
In the following sections we recast the assessment data gathered above using the conceptual framework described above.
Step One. Preliminary assessment of local position
A preliminary assessment of the local position was made using the same questionnaires that were mailed to the suppliers. This provided a mechanism for documenting the shortcomings of the current system and a basis for comparison between DYNIX and the alternative packages.
10 Current System
How well did the system meet the expressed needs of its users? DYNIX had allowed ITAM to automate many activities of the library that used to be manual. Students could, for the first time, consult information about the catalog on-line. The library had more automated control of the materials and was able to keep track of their circulation.
However, user satisfaction surveys showed that search mechanisms needed to be improved. Frequently, a search had to be done multiple times to find the materials wanted. All the queries had to be entered from terminals located in the library because a Web interface was not provided by the current system. Also, the system was very difficult to operate because its internal DMBS (Universe) provided limited functionality for accessing the data. Since the distributor provided no documentation about the internal data structure, the library staff had to guess where to locate a particular piece of information. Reports for decision support were produced after a long process of trial and error queries.
One of the most problematic issues was related to the execution of the periodic batch processes. Some of these processes, such as updating the basic indexes of the catalog, required that system use cease for an uncertain period of time. Regenerating the index of keywords, for example, took an average of 24 days in which the classification and cataloguing tasks had to be suspended. The system did not provide an estimate for the time needed to perform such tasks. Additionally, the lack of knowledge about the internal architecture of the system caused uncertainty about the effects that batch processes could have on the referential integrity of the data. When data were corrupted, the library had to load data from previous backups and to recapture the lost transactions manually.
How reliable and with what integrity did the system execute its information processing functions? The current system contained appropriate security mechanisms based on password protections for ensuring the quality of the information inputted to it. However, this scheme did not guarantee that referential integrity was preserved. Data could be directly input into the system using alternative procedures without going through the predefined screens. Since the data structures were unknown to the library staff, there was the danger of updating only some of the tables and losing the integrity of the data.
The system stored two types of data: passive and dynamic. Passive data were related to the bibliographical description of the material and had a high degree of reliability thanks to the check procedures performed by the library staff. Dynamic data were generated, modified or deleted by the procedures associated with the circulation of the materials when users borrowed or returned materials to the library. The reliability of these data was of a lesser degree. The most frequent errors with dynamic data occurred in a) the computation of sanctions, b) the deletion of users for no apparent reasons, c) inaccuracies in the statistics provided by the system, and d) the inconsistencies found in the inventories.
The current system also provided mechanisms for recovering data when a system crash occurred. These mechanisms worked well in general, but there were occasions in which data were lost or corrupted. Also, the system did not allow hiding information from the end-user. Everything that was captured could potentially be displayed. Fortunately, ITAM did not experience any problems caused by this limitation. On the other hand, some useful information could never be extracted because of lack of knowledge regarding the internal structure of the database.
11 Data inconsistency became a recurring problem that reduced the credibility of the current system and damaged the image of the library. Inconsistencies were caused by errors in the calculation procedures (e.g., when computing late penalties) or because not all the tables related to a particular transaction were properly updated. The library also experienced duplications in the information provided by the system. At the end of each semester, a file with the students officially enrolled for the next term, provided by the registrar, was used to determine who should be allowed to borrow materials from the library. On some occasions, student data were duplicated and manual debugging of the student file was needed. Less frequent were the duplications of bibliographical records having the same bar coding. These duplications were detected when the system reported a material as lost and the library staff noticed that the corresponding bar code was already used by another material.
The current system also presented problems related to the loss of data. Since these problems occurred regularly when computing late penalties, a second parallel system was developed and implemented. On one occasion, the library had to operate without a system for three weeks due to a major loss of data, and the library staff needed to make a full physical inventory to find which materials had been borrowed.
How well did the system contribute to the firm’s competence in its business processes? How did the system create value for the business customer? The system did not provide ITAM with any competitive advantage. It had become a “necessary evil” for the library staff due to the problems associated with its implementation. About 80% of the requests that ITAM made to the supplier were associated with making the system more usable. The remaining 20% were for solving problems with the current operation.
The system was only partially compliant with the MARC format v.19 and the ISO2709 norm for data transfers. Some fields were not defined with appropriate precision, and some subsets of this format were never implemented. Two major subsets that were missing were those relative to periodicals and to authorities by author and topic. Since the library was not compliant with these worldwide-accepted standards, its capability to cooperate with other libraries was negatively affected. Joint cataloguing projects, as well as the operation of exchange agreements, could not be established.
What resources were allocated to the system’s operation, use and maintenance? To what extent were they efficiently allocated? During the first years, ITAM paid $24,000 USD for an annual maintenance contract. However, due to the lack of adequate support from the distributor, ITAM preferred to work with him on a case-by-case basis. The library assigned a person to monitor to the operation of the system, and additional help from the computer center was obtained. Approximately 30 man-hours per week were dedicated to data maintenance, backups and development of additional reports.
The interaction of the end-users with the current system was inefficient due to the complexity associated with the search processes and the limited resources for querying the data. The library staff, however, learned to interact routinely with the system with acceptable efficiency. Several manual tasks were instrumented to “work around” the limitations of the package. For instance, in order to ensure that the balance of late fees was reliable, all the corresponding transactions were captured a second time in a separate system. A manual control was also kept for even the most basic statistics (e.g., the number of documents in the library), and contingency plans were regularly used because of the frequent periods in which the system was down.
12 The operation of the system presented many efficiency problems because only half of the modules were fully implemented. Also, the system provided a limited number of batch processes to facilitate updating large volumes of data. Therefore, most of the maintenance was performed on- line, record-by-record, through a time-consuming process. The system kept track of the table indexes that had to be regenerated and provided utilities for performing such tasks. Most maintenance tasks concentrated on improving the quality of the data or in correcting particular problems. No major maintenance to the system’s code had been performed prior to the assessment.
Future System
The assumption that the current system could receive proper maintenance and enhancement had already been shown to be wrong by the time the system was being considered for replacement. During the five years in which ITAM operated with DYNIX, the distributor of the package could not incorporate the complete MARC format into the system as requested. This capability would have allowed ITAM’s library to include the inventory of periodicals in the database, and to exchange reliable information with other libraries. However, at the time of the assessment, ITAM was the only institution in Mexico still using an ad hoc version that was only partially functional.
Some other requests made to the distributor that were never implemented included: a) fixing the routines for importing and exporting data for exchange purposes; b) developing an automated procedure that would permit identifying duplications in the data; c) developing procedures for updating the information in sets of records matching a particular criteria; and d) incorporating a tool for generating reports that could help in the debugging of the data stored in the system.
Table 2 summarizes the results of the preliminary assessment of the local position. This step of the method helped ITAM to document the limitations of the current system in a more formal manner and to identify the problems that had to be addressed by any replacement package.
13 Table 2: Preliminary Assessment of Local Position for ITAM’s Case Current System Future System (Assuming Maintenance and Enhancement of Current System) • Limited search and querying capabilities Not feasible: • Partial compliance with MARC format • ITAM had an ad hoc version of the • Slow batch processes package • No periodicals module • The distributor had not been able to fix • No Web interface the problems encountered or improve the • Frequent loss and duplication of functionality of the current system transactions • Unreliable dynamic data (sanctions, users, statistics, inventories) • Limited reporting • Hierarchical database • Limited batch processes for updating multiple records simultaneously
Step Two. Assessment of Industry's Leading Edge
In order to determine which solution would better fit its requirements, ITAM performed an assessment of the industry’s leading edge and common practice solutions using two sources of information. The first source was the questionnaires that the library sent to the corresponding distributors, and the second was the literature describing the major IOLS packages and their expected evolution that was consulted by the analysis group.
State-of-the-Art
Table 3 lists the principal commercial IOLS that were available in the second semester of 1998 when ITAM was doing the assessment (Bilal, Barry and Penniman 1999)7. This $ 475 million marketplace had dramatically changed during the previous two years due to joint-ventures and partnerships that had been established among the leading vendors. As a result, most systems exhibited hybrid characteristics and it was problematic to classify them in either the leading-edge or the global norming categories of the assessment grid. However, some experts considered the following characteristics to be in a state-of-the-art system:
• Full Compliance with the Z39.5 (v.3 1995) standard. The Z39.50 standard (Bibliotech Review 1999) was originally proposed in 1984 to provide a standard way of interrogating bibliographic databases. Since then, it has gone through three versions – in 1988 (v1), 1992=(v2) and 1995 (v3). Version 2 in 1992 also incorporated and became compatible with an ISO standard (10162/3) called Search and Retrieve. Version 3 in 1995 extended the features of the protocol and is the version that most suppliers are now implementing. However, many libraries were still using packages that did not fully support the standard. A state-of-the-art “Z-client” can send one or multiple requests simultaneously to several libraries and this
7 Numbers in parenthesis represent the total number of libraries still using each system and are included only for server-based systems.
14 feature allows substantial time savings. Also, Z39.50 opens up the exchange of bibliographic records among libraries.
• Object-Oriented Data Bases. OODB constitutes the leading-edge technology for information storage and retrieval, providing structural advantages such as modularity, inheritance and encapsulation. When the assessment was performed, TAOS was the only system built using an object-oriented approach (Evans 1999) and UCLA became the first major installation of this system.8
• Web OPAC. Leading-edge packages allow users to consult the catalogs using the Web. However, most vendors are still struggling with multiple OPACs for accessing their catalogues.
• Java-based Modules. The use of Java increases the ability of OPACs to offer more customized searching and more personalized interaction with the library system (Gallimore 1997, Chapter 11). Java also enables the code of the application to run on thin-clients, reducing the total cost of ownership of the library computing infrastructure (Bibliotech Review 1999).
• Compliance with the ISO 10160/61protocol. By the end of 1998 few commercial packages had implemented the ISO 10160/61 protocol for Inter Library Loans (ILL). Some examples of systems that provided such capability were WINGS from Pigasus and RELAIS from Relais International (Evans 1999).
• Compliance with the Z39.69 standard. One of the continuing problems with migrating to a new IOLS is the difficulty of moving the patron file, since many vendors store patron records in a proprietary format. A new standard for patron records, Z39.69, was developed for this purpose.
• Hypertext and Hypermedia. Some packages offer the possibility of navigating among documents on the OPAC by means of associative links (Duval and Main 1992). Unlike HTML, the EXtensible Markup Language (XML) is flexible and allows a home page developer to define the tags for any data items. A few vendors such as Endeavor and DRA are undertaking research projects involving XML encoded documents in their systems (Bilal, Barry and Penniman 1999). Also, some leading vendors are investing in systems that provide access to digital images, videos, audio, photographs, and documents.
• CD-ROM (and DVD) PACs integrated with the Circulation Module. Mayland’s public libraries were very active in networking CD-ROM Public Access Catalogs in order to provide users with access to CD-ROMs and DVDs. Leading-edge libraries have gone forward by
8 UCLA decided to upgrade its Orion library management system because it was not Y2K compliant and because its design, structure and scope were becoming obsolete. The development of ORION had begun in the 1960s and the system had been running in a mainframe since the early 1980s. After an extensive assessment process UCLA decided to purchase TAOS from Data Research Associates (DRA) for its state-of-the-art technology. By September 1999, TAOS had been beta-tested at a consortium of 40 academic libraries in Illinois. However, when the consortium postponed implementation, it put the UCLA library in the position of being the first very large installation of this system, something that UCLA “would not have elected to do” had it had a choice (UCLA 2000). Implementation at UCLA proved to be painful.
15 integrating the CD-ROM (and DVD) PACs with their circulation module in order to provide users with this information (Duval and Main 1992).
Table 3: Some Commercial IOLS Packages (Bilal, Barry and Penniman 1999) Library Vendor System Name ( Number of Installations) Ameritech Library Systems DYNIX (2,933), HORIZON (481), NOTIS (113) Best-Seller PORTFOLIO (160) Brodart PRECISION ONE, SUMMITONE, PINNACLEONE, LEPAC CARL Corporation CARL INFORMATION, MANAGEMENT AND DELIVERY SYSTEM 9 (IMDS) (36) , CASPR LIBRARYWORLD, COMPANION ALEXANDRIA DRA TAOS, CLASSIC, MULTILIS, INLEX (1,265) Endeavor VOYAGER (231) Ex Libris ALEPH 500 (418) Follett ALLIANCE PLUS, CATALOG PLUS, CIRCULATION PLUS Fretwell-Downing OLIB (134) Gateway LIBRARY MANAGEMENT SYSTEM (145) Gaylord GALAXY (308), POLARIS (9) Geac ADVANCE (227), PLUS (111), VUBIS (405) Inmagic INMAGIC DB/TEXT WORKS, DB/TEXT WEB PUBLISHER III, INNOPAC (888) International Library SYDNEYPLUS (84) Systems (ILS) Keystone KLAS (25) TLC LIBRARYSOLUTION (102) Nichols ATHENA Open Text BASIS TECHLIB (269) Sanderson SPYDUS LIBRARY MANAGEMENT SYSTEM (250) SIRS Mandarin SIRS MANDARIN Sirsi UNICORN (899) VTLS VIRTUA AND VTLS (317) Winnebago SPECTRUM, CIRC/CAT
Innovative Frontier
What will be the future of IOLS in the next years? This question is difficult to answer because libraries, as intermediaries, are dependent upon the actions of suppliers and users of information. However, libraries must attempt to predict the future to some extent if they are to plan beyond the short term. Gallimore (1997, Chapter 11) has predicted the following future scenario for libraries: • Increasing change, complexity and uncertainty. The speed of change in the Information Age is dramatic and libraries need to build dynamic and adaptive strategies to deal with this issue. Information is now being generated by multiple heterogeneous sources. Libraries will need to connect to more networks and to offer a wider range of services. Also, as few controls exist over the accuracy of information published on the Web, librarians are likely to play a major role in counteracting confusion and uncertainty.
9 CARL supports 757 libraries through its 36 installed systems.
16 • Increasing variety, portability, speed and integration of hardware platforms. Digital devices, wireless networks, portable clients, faster processors, plug-and-play technologies and multifunctional terminals are expected to transform the infrastructure of libraries in the near future. Faster and more powerful multi-user multi-threading open systems with parallel processing will be used to speed up transactions and queries (Yang 1998). • Increasing sophistication and integration of software. Push technologies, data-mining and expert systems are likely to revolutionize the manner in which a user interacts with an IOLS. Intelligent software will allow profiles of user preferences to be stored so that users can be alerted of new items of interest. IOLS will need to be better integrated with other software (e.g., ERPs) and resources of the organization. • Increasing bandwith and connections. Internet 2 will catalyze the development of a new breed of technologies for transmitting large volumes of digital materials through ubiquitous networks. • Increasing globalization, cooperation and dependence. Similar to what is occurring in multiple sectors of the economy, to remain competitive, libraries will need to establish partnerships with other libraries around the world. Standards will play a major role for enabling such type of collaboration. Libraries will increasingly find themselves in the role of gateways to external systems providing access to information over which they have no ownership rights and little control. • Increasing information, digitization, traffic and media ranges. Libraries will need to use effective ways for helping users navigate through huge volumes of available electronic information. DVD disks and new storage media will become widely used and will allow more digitization of materials. What technologies will be used in the digital library of the future? Some of the most exciting technology that is being developed and that is likely to be seen on the “Innovative Frontier” in future years is being investigated as part of the Digital Library Initiative (DLI) jointly funded by the U.S. National Science Foundation, the Department of Defense Advanced Research Projects Agency (DARPA) and the National Aeronautics and Space Administration (NASA). The Initiative's focus is to dramatically advance the means to collect, store, and organize information in digital forms, and make it available for searching, retrieval, and processing via communication networks, all in user-friendly ways. In the first phase of this initiative (1994-1998) six institutions were funded (NSF 1999): • The project at the University of Illinois at Urbana-Champaign had the goal of developing widely usable Web technology to effectively search technical documents on the Internet. Efforts concentrated on building an experimental testbed with tens of thousands of full-text journal articles from physics, engineering, and computer science, and making these articles available over the Web. Research focused on using the document structure to provide federated search across publisher collections.
• The INFORMEDIA-I project at Carnegie Mellon University combined speech recognition, image understanding and natural language processing technology to automatically transcribe, segment and index linear video. These same tools are applied to accomplish intelligent video search, navigation and selective retrieval. • The University of California at Berkeley digital library project concentrated on developing the tools and technologies to support highly improved models of the "scholarly information life cycle." The goal is to facilitate the move from the current centralized, discrete publishing
17 model, to a distributed, continuous, and self-publishing model, while still preserving the best aspects of the current model such as peer review. • The Alexandria Digital Library project at the University of California at Santa Barbara began in 1995 with the development of a working digital library with collections of geographically referenced materials and services for accessing those collections. • The University of Michigan digital library project had the goal of developing interfaces and infrastructures for users and providers such that intellectual work (finding, creating, and disseminating knowledge) is embedded in a persistent, structured context even while the underlying networked system is evolving. The core of the UMDL has been an architecture that supports the teaming of agents to provide complex services by combining limited individual capabilities. • Stanford University focused on interoperability issues and developed the “InfoBus” protocol which provides a uniform way to access a variety of services and information sources through "proxies" acting as interpreters between the InfoBus protocol and the native protocols. The impact of these projects in the evolution of commercial IOLS during the following years is still unclear. However, they are likely to modify the manner in which libraries operate in the future. For instance, during the execution of the DLI project at Stanford, the development team of found that the USMARC “wasn’t quite right” for the development of the new InfoBus protocol (Stanford 2000).
Table 4 summarizes the results of the assessment of the industry’s leading edge for ITAM’s case. During future years, it is likely that some of the characteristics of IOLS that are now considered to be part of a state-of-the-art system will become part of a common practice solution. Also, it is probable that new issues will emerge on the innovative frontier, depending upon the evolution of the role played by libraries.
Table 4: Assessment of Industry’s Leading Edge for ITAM’s Case State-of-the-Art-System Innovative Frontier • Full compliance with Z39.5 standard for • Full digital access via the Web to large bibliographic databases collections of digital materials • Full compliance with the 10160/61 • Speech recognition, image understanding protocol for interlibrary loans and natural language processing • Full compliance with the Z39.69 standard • Distributed and continuous self- for patron records publishing from multiple sources • Object-Oriented database • Services for accessing geographically • Web OPAC referenced material • Java-based modules • Interoperability among diverse services • Hypertext and Hypermedia and information sources • Integration of Circulation Module with CD-ROM and DVD Public Access Catalogs
18 Step Three. Assessment of Global Norming
Common Practice
The members of the Association of Research Libraries (ARL) represent the high end of the common practice of library automation. To serve these libraries, a library automation system must be mature and include a full suite of modules for each aspect of library operation. Also, the system should be able to handle extremely large collections and sustain many multiple users and a high volume of transactions. Table 5 shows the IOLS being used by each of the ARL members when the assessment was performed. As noted by Breeding: “Many of these libraries are migrating to second or third generation systems. Those that implemented mainframe based systems such as NOTIS and GLIS have or soon will migrate to client-server systems” (Breeding 2000).
19 Table 5: IOLS used by ARL Libraries (Breeding 2000) No. of ARL System ARL Libraries Libraries Arizona, Arizona State, Boston University, Brown, CISTI, University of California - Irvine, University of California - Riverside, University of California - San Diego, Case Western, CRL, Colorado, Cincinnati, Colorado State, Dartmouth, Georgetown, Houston, Kent State, INNOPAC 33 Massachusetts, Miami, Michigan State, Missouri--Columbia, Nebraska, New Mexico, New York Pubic Library, Ohio State, Ohio University, Oregon, Saskatchewan, Temple, Washington, Western Ontario, Washington State, Washington University -- St. Louis. Alabama, Auburn, Connecticut, Cornell, Library of Congress, George Washington, Georgia, Georgia Tech (by late 2000), Guelph, Hawaii, VOYAGER 26 Kansas, Kentucky, National Library of Medicine, Northwestern, Oklahoma State, Pittsburgh, Princeton, Purdue, Pennsylvania, Queen's, Rochester, Syracuse, Texas A&M, Tulane, Waterloo, Wisconsin. Brigham Young, Emory, Indiana, New York, Oklahoma, Penn State, UNICORN 13 Rice, Rutgers, Stanford, Southern California, Virginia, Vanderbilt, York. Columbia, Delaware, Florida, Florida State, Harvard, Howard, Louisiana NOTIS 12 State, Michigan, Minnesota, South Carolina, University of California - Santa Barbara, Yale. Boston Public, British Columbia, Duke University, North Carolina State, Southern Illinois, Texas Tech, University of Alberta, University of DRA 12 California - Davis, Illinois--Urbana-Champaign, Manitoba, North Carolina -- Chapel Hill, Toronto. Chicago, Iowa State, Illinois -- Chicago, Johns Hopkins, Linda Hall, HORIZON 10 McMaster, Smithsonian Institution, Tennessee, Utah, Wayne State. Iowa, McGill, Notre Dame, State University of New York -- Albany, ALEPH 6 State University of New York at Buffalo, State University of New York - - Stony Brook. TAOS 1 University of California - Los Angeles. ADVANCE 2 MIT, New York. VTLS 2 National Agriculture Library, Virginia Tech. CARL 1 Maryland. MELVYL 1 University of California - Berkeley AMICUS 1 National Library of Canada MULTILIS 1 Universite Laval LOCALLY University of Texas -- Austin 1 DEVELOPED TOTAL 122
The majority of commercial library management systems being used by the bulk of libraries in 1998 had the following characteristics (Cibbarelli 1995):
• Acquisitions. Commercial IOLS provide access to information about materials that have been ordered but have not yet been received and processed for circulation.
20 • Cataloguing. The system is compliant with the MARC II format and allows importing and exporting full MARC records. Multiple authority files are already setup. Users have online access to the catalog for querying by multiple fields. The package generates listings of items in the collection in multiple formats and media (i.e., microfiche, paper, cards, CD-ROMs).
• Multiple Format of Bibliographies. Citations can be formatted using a variety of standard formats such as the MLA Style Sheet, the Chicago Manual of Style and Tarabian (Cibbarelli 1995, p.22). • Inter-Library Loans. Most systems provided mechanisms for loaning materials form one collection to another agency. • Index of Keywords. Similar to a thesaurus, an index provides an alphabetic listing of all terms used in a field.
In addition, the following characteristics were becoming a common practice among IOLS (Bilal, Barry and Penniman 1999): • Multiple Search Strategies. Most packages provided the end-user with means for performing complex queries through different search strategies. Also, they let the user save the results of a search into a file or send them by e-mail.
• GraphicalUser Interfaces. From a recent survey, more than half of the vendors mentioned that all their modules were either already developed for or were converted to Windows in 1998. As a result, Graphical User Interfaces (GUIs) have become the norm and Windows-based conversions are predominant.
• Relational Databases. Vendors are abandoning proprietary databases and migrating their applications to well established commercial databases such as Oracle. This provided the user with mechanisms for querying the data and for linking the IOLS with other systems of the organization.
In a survey of IOLS that covered not only functionality of an IOLS but also other issues related to its implementation, Cibarrelli (1993) found: • Most packages have poor documentation. • Training is the category with the lowest average rating. • Service and support are key to the long-term user approval of library automation software. Geography still plays an important role in product support. • Vendors are doing a good job of creating reliable software with appropriate capabilities. • Ease of use is important since generally the package will be used not only by library staff but also by others.
In Mexico the situation was not very different when ITAM analyzed the replacement of DYNIX. Most major libraries in Mexico had rapidly moved towards automating their processes by means of acquiring a leading commercial package with a distributor in the country and lagging institutions were trying to catch up by investing in already proven solutions. However, only those automation projects that were led by librarians with good knowledge about information systems had provided the expected results.
21 Business Convergence
In the future, only those packages that are fully compliant with the NISO and ISO standards are likely to remain in the market. Full compliance with the Z39.5 is rapidly becoming mandatory for survival and the implications of this change will be very significant. Z39.50 allows extremely powerful search statements to be defined including complex Boolean statements involving all the standard operators AND, OR and NOT, comparison operators for dates, proximity searching, truncation and completeness (Bibliotech Review 1999). Also, the standard enables a) authentication, so the Z-server can control who accesses their databases; b) accounting/resource control, to allow access to be charged for; c) index browsing as typically available in OPAC systems; e) definition of new record formats (such as MARC); and e) facility to explain remote database services.
A similar situation will occur with respect to the other standards developed for IOLS such as the ISO10160/61 protocol and the other Z39 standards (e.g., Z39.69). Since Inter-Library Loans will need to be efficiently instrumented by the IOLS, compliance with the ISO 10160/61protocol will be necessary.
In addition, not just the OPAC will need to operate on the Internet, but also other modules of the IOLS. This is a tendency that has been observed in other integrated applications for organizations (e.g., ERP´s) and is likely to permeate to IOLS.
Java and XML will be used in the newer versions of the commercial IOLS running under thin clients. Applications will take advantage of a three-tier architecture supported in transaction monitors such as Tuxedo. Network PCs or network computers will replace most of the PCs used as clients in libraries.
Table 6 summarizes the results of step three of the method for ITAM’s case. Note that some of the characteristics of a state-of-the-art system such as compliance with the new standards of the industry are expected to become a common element of most commercial packages within the next three (instead of five to be consistent) years.
Table 6: Assessment of Global Norming for ITAM’s Case Common Practice Business Covergence • Full compliance with MARC II format • Full compliance with the Z39.5 standard • Support of multiple search strategies and for bibliographic databases complex queries • Full compliance with the 10160/61 • Support for multiple bibliography formats protocol for interlibrary loans • Mechanisms for tracking materials in • Full compliance with the Z39.69 standard transit for patron records • Support for interlibrary loan transactions • Web OPAC and other modules in Internet • • Graphical User Interface Three-tier architecture • Java and XML modules • Relational database
22 Step Four. Reassessment of Local Position
ITAM prepared a comparative report of the alternative packages and the current position based on a comprehensive list of requirements for each of the IOLS modules. Priority was given to those items that were considered to be essential (i.e., a “common practice”) for the proper operation of the library. Then, those elements of a state-of-the-art system that were expected to become part of global norming were evaluated.
Additionally, the following criteria played a major role in the evaluation:
• Local Distributor. ITAM limited the search to those packages that were available by means of a local distributor. This decision was taken because receiving technical support on-site was considered a top priority. As a result, only the following systems were analyzed: ALEPH 500, INMAGIC DB/TEXTWORKS, HOLLIS, INNOPAC, LIBRARY 4 UNIVERSAL (L4U), SIDNEY PLUS LIBRARY MANAGEMENT V.3.3, UNICORN and VOYAGER. The evaluation team had many meetings with the local distributors of the packages and visited some of their customers in order to have a better idea of the quality of the service that would be received from each of them.
• Development Team. The selected package had to be supported by a strong development team that would take into consideration future trends in hardware and software at the innovative frontier. Interviews with some of the technicians who maintain and enhance the commercial packages were also undertaken.
• Training. As with other system implementation projects, the time required to customize a package (in terms of setting local parameters) depends upon many technical and organizational elements. A key success factor for implementation has been the knowledge that the library staff acquires during the training period. Training materials were analyzed in order to determine to what extent they would provide the staff with the abilities required for searching, recovering and backing up the information, as well as for updating the records of the databases.
After doing the assessment for both the leading-edge applications as well as for those used in the common practice, the decision to replace DYNIX became even more convincing. Not only was ITAM suffering from the problems with the “current system” described in Step One, but the gap with other libraries was widening.
Some conclusions of the assessment regarding the benefits that would be obtained by using a new commercial package that complies with the requirements of ITAM were the following: a) Effectiveness. For end-users, the major accomplishment of the new package would be to improve the search for bibliographical materials and to provide the capability of performing transactions by other means. With the new system, end-users would have remote-access to the catalog and could save the results of their search procedures.
From the point of view of the library staff, the selected package would need to offer the tools required for interlibrary exchange and collaboration, as it would be compliant with the MARC format, and with the ISO2709 and Z39.50 standards. Additionally, the system would provide mechanisms to ensure the reliability of dynamic data as well as reports for decision support.
23 Finally, the system would allow the user to generate customized reports directly or with the help of external software tools. b) Efficiency. ITAM expected to have some gains in efficiency due to the elimination of some of the additional work that was needed to keep the three modules of the current system running. However, it did not expect to cut people because the staff would be providing not only the tasks supported by the current system, but also the additional services that so far had not been implemented. The most expensive resources were those associated with the computing infrastructure required for running the system, and with the personnel assigned to the installation, support and maintenance of the package.
Also, ITAM expected to perform creation and update of the library databases on-line and reduce the execution of batch processes to a minimum, leaving those processes for exchanging information with other libraries. The most resource consuming maintenance tasks would be those related to the implementation of new software releases. However, it was expected that the cost of these tasks would be less than the amount now being paid for correcting the problems with the DYNIX system. c) Confidentiality. The new system would provide better mechanisms for protection of sensitive information from unauthorized disclosure. Sensitive information such as sanctions would be kept hidden from regular users. d) Integrity. The new system would solve many of the problems currently encountered with data integrity. Data duplication and referential integrity would be facilitated by a more robust data structure. e) Availability. During stabilization of the new system, some interruptions were expected to occur. An average of two or three interruptions per month is an acceptable benchmark. After this period, the operation of the system should only be interrupted when it is mandatory for the purpose of performing full back-ups of the information. f) Compliance. The new system would be compliant with the latest version of the MARC format, including the complete hierarchies of authorities. Also, it would comply with the ISO2790 and the Z39.50 standard for information exchange. This would allow the library to incorporate into the database MARC records produced by other cataloguing agents around the world. g) Reliability of Information. ITAM expects the new system to be much more reliable than the current hybrid version of DYNIX for several reasons. First, ITAM would select only a system that had already been tested successfully in other libraries in Latin America. Second, the experience that ITAM library had gained by having to struggle with the DYNIX system would become an asset when trying to implement a new system, particularly if this is a descendant of the current system.
Some of the additional functionality that would be obtained with the help of a new package that complies with ITAM’s requirements included the following: • ITAM would be able to use the Internet as a distribution channel for services provided to the community. This would allow ITAM to gain presence in this important medium and reduce the risk of being left isolated.
24 • End-users could search for materials and interact with the library processes using the Web and download data as needed. This capability would reduce the loads of the central servers. • The Internet could also be used by the library staff to improve its analysis tasks and the normalization of the description of the different types of materials that belong to the library. • The staff of the library could better exploit the information stored in the system in order to create ad hoc reports not provided routinely. • The library could automate the collection and management of periodicals that were handled manually. • ITAM would have better control of the reserves, borrowings, returns and sanctions. • The library could provide the end-user with more friendly and effective tools for searching among the different collections. • Library staff would be in a better position to customize the system according to different user profiles. Also, they would be able to import information to the system without affecting the referential integrity of the data.
Table 7 summarizes the results of Step Four of the method for ITAM’s case. The table lists the desired characteristics of a replacement package including some shortcomings of the current system that had not been identified in Step 1 (e.g., compliance with the Z39.69 standard) and that were the result of the analysis performed in Steps 2 and 3. In some elements, ITAM was flexible on choosing either a state-of-the-art solution or a global norming characteristic (e.g., a relational or object-oriented database).
25 Table 7: Reassessment of Local Position for ITAM’s Case Current System Required System • Limited search and querying capabilities • Powerful search and querying • Partial compliance with MARC format • Full compliance with MARC II format and the Z39.5 standard for bibliographic databases • Slow batch processes • On-line processing; reduce batch • Limited batch processes for updating processes to a minimum: automated batch multiple records simultaneously process for multiple record updating • Text-based Interface • Graphical User Interface • No periodicals module • Include cataloguing, circulation, • Frequent loss and duplication of acquisitions, periodicals and OPAC transactions modules
• Unreliable dynamic data (sanctions, users, • Reliable data; avoid recapture and statistics, inventories) duplication; maintain referential integrity • Unable to process interlibrary loan • Full compliance with the 10160/61 transactions protocol for interlibrary loans • Noncompliant with the Z39.69 standard • Full compliance with the Z39.69 standard for patron records for patron records • No Web interface • Web OPAC and other modules in Internet • Limited reporting • Flexible ad hoc reporting • Non SQL database • Relational or Object-Oriented database • One-tier architecture • Three-tier architecture • Centralized proprietary code • Java and XML modules Others: • Local distributor • Strong development team • Adequate training
Step Five. Reestablishment of Local Position
Given the absence of a broader portfolio context in ITAM’s case, the last step of the assessment consisted of deciding upon the preferred replacement package, taking all factors into account. ITAM did not want to buy a system that would have support problems in Mexico. As a consequence, the direct participation of the developer in the installation and customization of the system was considered a top priority in the evaluation of the package.
After some additional interviews with the finalist vendors and with librarians from leading user organizations, ITAM decided upon Ex Libris’s ALEPH 500 package. This system offered the following characteristics:
• ALEPH 500 is a complete, integrated system that manages all aspects of the library including Web and GUI PAC, Circulation, Cataloging, Serials, Acquisitions and ILL.
26 • The system’s architecture is open and modular, making it customizable and scalable. It may be easily interfaced with other systems and databases.
• A multi-tier client/server architecture relying on an Oracle database combined with an open- system, API-based design reflects the use of state-of-the-art technologies also aligned for likely business convergence.
• The system had recently been acquired by several institutions in Latin America similar to ITAM, such as the Universidad Católica de Chile, which in the early 1990's was also one of the first Latin American universities to implement DYNIX.
• ALEPH 500's multilingual, multiscript, and multidirectional text capabilities handle content and interface in 20 languages and many scripts.
• Even though it is just beginning to penetrate the ARL libraries, ALEPH 500 has a considerable number of world-wide implementations as shown in Table 3.
With regard to formal approval, the investment project for replacing the library management system had not been included in ITAM’s 1999 IS portfolio. However, the assessment performed by the task-force team helped to convince the authorities of ITAM to allocate additional resources for this project.
Implementation Postscript
Implementation of the ALEPH 5000 package began in May 1999. Version 505.11.4 of ALEPH was installed in an Alpha Server 4000 with two 300 Mhtz processors and 1 Gb of RAM running Unix 4.0E with Oracle 7.3.3. The contract that was signed with the distributor of ALEPH was remarkably different from the one that was signed previously for the implementation of DYNIX. The direct participation of knowledgeable people in the definition of requirements resulted in a contract that is very detailed in terms of the specifications for the new system. The distributor was obliged to provide training and continuous support to the staff in order to ensure the proper implementation of the package. The first training course for the system was given to library staff during the first two weeks of June 1999.
During the months of June and July 1999, the library customized the system by defining the following: • Structure of the different types of cataloguing records • Types of bibliographic materials • Collections • Types and length of loan periods • Calendars and working hours of the library • Types and privileges of different user profiles • Field and subfields of the MARC format • Indexes
ITAM started the conversion of bibliographic records to the MARC standard in February 1999, even before the decision upon the replacement system had been taken. An external firm was hired to extract and normalize the records stored in DYNIX to the MARC II format. Then, during the
27 period of June and July 1999, these records were transferred to ALEPH. By the end of July 1999, 67,237 bibliographic records and 105,540 volume records had been converted.
Loading of user records was accomplished in two periods (July 1999 and December 1999) using data provided by the Banner2000 scholar system of ITAM (SCT 2000). Cataloguing and Web OPAC were the first modules to be implemented. With these modules, students and faculty were able for the first time to perform multiple search and complex queries using the Web.
Four point-of-sale printers were bought in order to have a better control of loans by printing a receipt for every transaction made in the Circulation module. Also, several automated routines were developed in order to send e-mail messages to those users that are late in returning the materials or that have made a reserve on a particular reference.
To complement the reporting capabilities provided by ALEPH, in November 1999 ITAM bought the Discover 2000 tool. This tool provides the users with mechanisms for extracting data from the Oracle tables of the system in a user-friendly manner.
Current Situation
ITAM has been able to implement most of the functionality of ALEPH in a short time. During the implementation process, ITAM reported to the distributor, Sistemas Lógicos, and the developer, Ex-Libris, every issue that it found regarding the functionality of the corresponding modules. These issues, as well as three minor problems encountered during the Y2K transition were resolved promptly in a satisfactory manner.
Currently, ITAM is developing SQL scripts to automate the generation of several indicators that will help to monitor the quality of the services provided by the library. Also, some messages of the Circulation module are being translated into Spanish. With respect to Cataloguing, ITAM is modifying the database of Authorities in order to account for the MARC format and the internal operation policies of the library.
Lessons from the Exercise
Our analysis of how ITAM could have framed its decision to replace the DYNIX system using the proposed method provides us with some lessons regarding the applicability of this framework for other assessment processes.
First, the distinction between leading-edge and global norming might be fuzzy for some types of applications. More specifically, where commercial packages dominate, the current state-of-the-art may blend rather closely with business convergence, as vendors struggle with their products to meet the new offerings of competitors. For instance, in the case of IOLS some features such as the level of compliance with the Z39.50 standard could have been considered to belong to either of these categories (and thus we listed them for both).
Potentially, a similar situation might have arisen in distinguishing between operational and developmental features, in particular at the leading edge, where announced package features might not yet be successfully implemented with customers, for instance (as with “vaporware”). However, we encountered no such example in the present case. In any event, neither of these
28 classification situations poses a serious problem for applying the method. In both cases the assessment grid would still help us frame the system replacement decision by bringing external and internal elements into consideration. 10
Second, the method does not necessarily imply a linear execution of its different steps. In particular, Steps Two and Three can sometimes be usefully carried out in parallel, especially where commercial packages dominate the solution options. In the ITAM case, we looked at leading-edge and global norming solutions in parallel since a priori there was no clear distinction as to which application features belonged in either category.
Third, the assessment must be careful to incorporate not only technical but also organizational elements into consideration. For instance, where commercial packages are involved, the effectiveness of the support and training that may be received from each vendor might play a significant role in the evaluation, as was the case for ITAM. More broadly, then, the notion of the future system should incorporate more than just the application software and its features. It should include closely associated work activities.
5 Conclusion
The present paper offers a straightforward and we hope practical approach to framing the system replacement decision within the firm. It brings together both “internal' and “external” logic in assessing whether an application system may have reached the end of its useful life. In particular, it sensitizes the assessment to innovation in the firm's competitive environment. Several cautionary notes to our proposal are nevertheless in order.
First, further research in the form of trial and experimentation with the approach in actual organizational settings is obviously required to confirm the approach's usefulness and to suggest implementation guidelines. Here we have merely sketched the proposed method and provided one in-depth illustration of its applicability.
Second, while we have attempted to help frame the system replacement decision, we have presented no specific model for carrying it out. We have offered no integrative cost-benefit structure within which the “best” decision might be reached. Our limited aim here has been simply to provide the contextual frame within which the alternatives deserving of such analysis can be identified.
Acknowledgments
The authors want to acknowledge the help of Catalina Jaime, Director of the Library of ITAM, for providing us with information for the illustrative Case Study of Section 4.
10 See also our discussion of “waves of innovation” in Appendix I. Making distinctions among the innovation categories is not necessarily expected to be clear-cut.
29 References
Abrahamson, E. and Rosenkopt, L. (1993) Institutional and Competitive Bandwagons. Academy of Management Review, Vol. 18, 487-517 Bancroft, N., Seip, F.L., and Sprengel, A. (1998) Implementing SAP R/3. Second Edition, Greenwich, CT.: Manning Publications Baroudi, J.J. and Orlikowski, W.J. (1988) A Short-form Measure of User Information Satisfaction. Journal of Management Information Systems, Vol. 4, No. 4, 44-59 Barua, A. and Mukhopadhyay, T. (1989) A Cost Analysis of the Software Dilemma: to Maintain or to Replace In Proceedings of the 22nd Hawaii International Conference on Systems Sciences Bibliotech Review (1999) Java, Network Computers and Network PCs, http://www.biblio- tech.com/html/java__ncs_and_netpcs.html, Accessed: February 12, 2000 Bibliotech Review (1999) The Z39.50 Standard, http://www.biblio-tech.com/html/z39_50.html, Accessed: February 12, 2000 Bilal, D., Barry, J., and Penniman, W.D. (1999) Automated System Marketplace 99: A Balancing Act. Library Journal Digital, No. April 1 Brady, T., Tierney, M., and Williams, R. (1992) The Commodification of Industry Applications Software. Industrial and Corporate Change, Vol. 1, 489-513 Breeding, M. (2000) Automation Trends: ARL Libraries, http://staffweb.library.vanderbilt.edu/breeding/arl-systems.html, Accessed: April 18, 2000 Chan, T., Chung, S.L., and Ho, T.H. (1994) Timing of Software Replacement In Proceedings of the l5th International Conference on Information Systems Vancouver, Canada, 291-307 Chan, T., Chung, S.L., and Ho, T.H. (1996) An Economic Model to Estimate Software Rewriting and Replacement Times. IEEE Transactions on Software Engineering, Vol. 22, No. 8, 580-598 Cibbarelli, P.R. (1993) User Ratings of Library Automation Software: How Satisfied are Information Professionals with the Software Products They are Using? In Proceedings of the Integrated Online Library Systems IOLS-93. Cibbarelli, P.R. and Nixon, C., eds. New York: Learned Information, Inc., 33-45 Cibbarelli, P.R. (1995) IOLS Software: Options for School Libraries and Media Centers In Proceedings of the Integrated Online Library Systems IOLS-95. Cibbarelli, P.R. and Nixon, C., eds. New York: Learned Information, Inc., 21-33 Cibbarelli, P.R. and Cibbarelli, S. (1998) Directory of Library Automation Sofware, Systems and Services, Medford, NJ.: Information Today, Inc. Dimaggio, P.J. and Powell, W.W. (1983) The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields. American Sociological Review, Vol. 48, 147-160 Duval, B.K. and Main, L. (1992) Automated Library Systems: A Librarian's Guide and Teaching Manual, Westport, U.K.: Meckler Espinoza, L.E. (1988) Sistema de Publicaciones Periódicas para la Biblioteca Raúl Baillères del ITAM, Tesis de la Ingeniería en Computación, Instituto Tecnológico Autónomo de México: México, D.F. Evans, P.W. (1999) ALA's Midwinter Show: Products, http://www.biblio-tech.com/html/ala_midwinter_99.html, Accessed: February 15, 2000 Gallimore, A. (1997) Developing an IT Strategy for your Library, London: Library Association Publishing
30 Gamboa-Hirales, C.I. and Izkowich, R. (1990) Sistema de Control de Adquisiciones de Material Bibliográfico, Tesis de la Ingeniería en Computación, Instituto Tecnológico Autónomo de México: México, D.F. Genaway, D.C. (1984) Integrated Online Library Systems: Principles, Planning and Implementation, Whitle Plains, NY.: Knowledge Industry Publications Gode, D.K., Barua, A., and Mukhopadhyay, T. (1990) On the economics of the Software Replacement Problem In Proceedings of the 11th International Conference on Information Systems Copenhagen, 159-170 Hanna, M. (1993) Maintenance Burden Begging for Remedy. Software Magazine, Vol. 13, No. 6, 53-63 Joseph, T. and Swanson, E.B. (1998) The Package Alternative in System Replacement: Evidence for Innovation Convergence, in Information Systems and Technology Innovation and Diffusion, Larsen, T.J. and McGuire, G., Editors, Idea Group Publishing: Harrisburg, PA. (1998), 375-389 Lieber, R.B. (1995) Here comes SAP. Fortune, No. October 2, 122-124 Lientz, B.P. and Swanson, E.B. (1980) Software Maintenance Management, Reading, MA.: AddisonWesley Nolan, R.L. (1982) Managing Information Systems by Committee. Harvard Business Review, No. July/August, 72-79 NSF (1999) Digital Libraries Initiative, http://www.dli2.nsf.gov/dlione/, Accessed: January 20, 2000 Ramiller, N.C. (1996) Airline Magazine Syndrome: A Corporate Parable and its Lessons for Executives, IS Working Paper 3-96, The Anderson School at UCLA: Los Angeles, CA. Rogers, E.M. (1983) Diffusion of Innovations. Third Edition, New York: The Free Press SCT (2000) Banner2000, http://www.sctcorp.com/industrysolutions/education/banner2000S.htm, Accessed: April 18, 2000 Stanford (2000) The Stanford Digital Libraries Project, http://walrus.stanford.edu/diglib/, Accessed: February 15, 2000 Swanson, E.B. (1995) Findings from an Information System Census, Information Systems Working Paper #8-95, The Anderson School at UCLA: Los Angeles, CA. Swanson, E.B. and Beath, C.M. (1989) Maintaining Information Systems in Organizations, New York: Wiley Swanson, E.B. and Ramiller, N.C. (1997) The Organizing Vision in Information Systems Innovation. Organization Science, Vol. 8, No. 5, 458-474 UCLA (2000) Frequently Asked Questions about ORION2 and the Implementation Process, http://orion2.library.ucla.edu/o2faqs.htm, Accessed: April 17, 2000 Xenakis, J.J. (1996) Taming SAP. CFO, No. March, 23-30 Yang, S. (1998) Newer Operating Systems: Analysis of their Utilization and Potential in Libraries In Proceedings of the Integrated Online Library Systems IOLS-98. Cibbarelli, P.R. and Nixon, C., eds. New York: Information Today, Inc., 151-156
Appendix I. Background in Innovation Theory
The process of innovation adoption among organizations is typically characterized by an S-shaped diffusion curve when the cumulative number or percentage of adopters is plotted over time (Rogers 1983). Early adoption is slow, but the rate increases gradually before leveling off and then declining with late adoption. The adopters themselves are accordingly often classified by their temporal position in this process. As a population, they tend to be distributed normally around the
31 average time of adoption. The true innovators are sometimes said to be the first 2.5% of adopters. The early adopters are the next 13.5%. These are followed by the early majority, the next 34%, bringing the total to half the adopters. As the rate of adoption now drops off, the late majority constitutes another 34%. The laggards, the last 16%, complete the distribution. (See (Rogers 1983, Chapter 7)). Non-adopters are ignored in this characterization.
In the present paper, we posit waves of innovation in a business application within an industry. At any point of time we may be able to observe: (1) the innovative frontier occupied by the current innovators, (2) the current state-of-the-art being practiced by the early adopters, (3) the business convergence now reflected in the adoption decisions of the early majority, and (4) the common practice to which the late majority if not yet the laggards has now committed itself. We have accordingly devised the Application Assessment Grid to direct managerial attention to each of these potential waves. The extent to which each wave breaks sharply from the preceding one(s) is of course likely to vary both over time and from application to application. Innovation is hardly likely to be an orderly and predictable phenomenon, especially across domains. Thus, in applying the AAG from one application domain to another, very different innovation characteristics may be identified.
Appendix II. Background in Integrated Online Library Systems
An IOLS can be defined as “a library system that uses a common machine-readable database and has two or more subsystems operational and accessible online” (Genaway 1984, p.4). IOLS emerged due to the adoption of the MARC (Machine Readable Catalogue) standard developed by the Library of Congress in the late sixties for cataloguing materials using variable length records. The MARC II format that was adopted in 1974 extended the original scope of the standard to account for other types of non-monograph materials. Later, this standard became the basis of the Z.39.5 standard developed by the National Information Standards Organization (NISO) (Bibliotech Review 1999) that is used for automating library transactions.
Most commercial IOLS have five core modules (Duval and Main 1992): • Cataloguing. This module is the basis of the IOLS because it stores the bibliographic records using the MARC II format. Authority files consisting of unique name and subject headings are established to provide a single form of entry for a particular entity and to reduce the possibility of ambiguity between titles or concepts that are similar. Examples of name headings are authors, corporate bodies, conference names and series, among others. Authority records provide the system with cross-references among forms of entry and relations to other references that are related in some way to a given heading. The cataloguing module should allow to create records from automated cataloguing sources such as the Online Computer Library Center (OCLC). • Circulation. The circulation module is used to process the activities related with the lending and return of bibliographic materials. This module stores a list of the patrons (i.e., library users) that may borrow materials, the parameters used to compute late fines and the rules that are used for lending, reserving and renewing materials. • Acquisitions. The acquisition module is the link between the library and its vendors according to the standards developed by the Book Industry Systems Advisory Committee (BISAC) for this purpose. It provides support for the purchasing process, including the handling of new
32 and standing orders, blanket orders, subscriptions and approval plans. This module manages the multiple funds that are used for acquiring the materials. • Serials. This module is used to handle all types of serials (e.g., periodicals, journals, newspapers, government documents, etc.). Acquisitions of serials are automated using this module by sharing some information about funds and vendors with the acquisitions module. • Online Public Access Catalog. The Online Public Access Catalog (OPAC) is the component of the system that lets the user browse through the catalog materials and perform queries for finding references. The first commercial OPACs appeared when several universities joined the OCLC network that originated in Ohio State University during the mid and late 1970s. Other institutions, such as the University of California have developed their own public access catalog (MELVYL). These five modules can be organized into three types of architectures (Genaway 1984, Chapter 1): a) Non-redundant single file configuration, in which a single copy of full MARC records is stored centrally; b) Redundant multiple file configuration, in which several files are created from the master bibliographic record store for different subject headings to speed up the searches; and c) Non-redundant multiple file configuration, in which the master MARC records are split into its main components or parts and linkage is provided by means of control numbers.
Perspectives for an Assessment
The overall functionality of an IOLS can be evaluated from three different points of view. The point-of-view of the end-user is determined by how easy he or she perceives that it is to consult the inventory of bibliographical materials and by the effectiveness and efficiency of the services that they receive from the library. These services include borrowing, returning and reserving materials, proper calculation of sanctions for late returns, recovering and manipulating data, and general interaction with the system. The two main components of an IOLS related to these processes are the circulation and the cataloguing modules.
The point-of-view of the library staff is determined by the capabilities of the system for feeding and manipulating bibliographical records as well as for supporting the services provided to the community. The library staff people routinely interact with the system for performing their work and constitute heavy users of the modules for cataloguing, acquisitions, periodicals and circulation.
Finally, the computing people responsible for keeping the system running are mostly concerned with the capabilities provided to perform backups, recover data, implement maintenance tasks and guarantee the security of the system.
33