Desktop Information Delivery for Effective Administration: Client/Server Solutions Copyright CAUSE 1994. This paper was presented at the 1993 CAUSE Annual Conference held in San Diego, California, December 7-10, and is part of the conference proceedings published by CAUSE. Permission to copy or disseminate all or part of this material is granted provided that the copies are not made or distributed for commercial advantage, that the CAUSE copyright notice and the title and authors of the publication and its date appear, and that notice is given that copying is by permission of CAUSE, the association for managing and using information technology in higher education. To copy or disseminate otherwise, or to republish in any form, requires written permission from CAUSE. For further information: CAUSE, 4840 Pearl East Circle, Suite 302E, Boulder, CO 80301; 303449-4430; e-mail info@cause.colorado.edu Desktop Information Delivery for Effective Administration Client/Server Solutions Gary M. Hammon, Manager University Information Resources University of Massachusetts The President's Office Amherst, Massachusetts ABSTRACT Current and proposed changes in the financing and management of higher education are fueling an ever increasing demand for information at the desktop of administrators and executives. Client/server solutions can deliver critical information while enabling the IT organization and administrative customers to maximize computer resources, and more importantly, limited human resources. This paper provides guidelines and strategies for implementing client/server solutions for management reporting, decision support, and executive information systems. A practical framework for using client/server architecture will be discussed with examples of applications implemented in a complex multi-vendor environment at the University of Massachusetts. INTRODUCTION CAUSE 93 comes at a time when the need for information to manage and support the mission of our institutions has never been greater. Regulators, legislatures and the public expect clear information describing the effective operation and value provided by each institution. This was highlighted in the June 1993 SHEEO/NCES Network Conference (State Higher Education Executive Officers/National Center for Education Statistics) which discussed important changes at NCES. "These changes have resulted from several causes including technological developments, the emergence of additional federal and state data needs and reporting requirements, and organizational changes within NCES and other agencies."(SHEEO/NCES,1993) As Information Technology continues to evolve, there are more options available to provide the management information that is in such high demand. This paper provides guidelines and strategies for implementing one such option, client/server applications, as a solution for management reporting, decision support and executive information systems. The paper concentrates on the business issues and success factors rather than any specific technical designs. THE CHALLENGE Much has been written of the rate of change in information technology, especially in view of the ability of systems to offer meaningful information for management of the enterprises. Without regard to the debate of evolutionary change versus revolutionary change, we need to take stock of the situation at our institutions and identify rapid deployment strategies to deliver the critical information needed today, (or more likely in our customers eyes - yesterday). While our strategies should also have an eye to the future we must avoid the inclination to design the perfect system - time is of the essence. THE CHANGES The 1980's saw extensive growth in our legacy systems and data stores. Disk storage costs decreased, new programmer productivity tools and methodologies became more effective, and we began to see data as a resource to be managed! Of course like all valuable resources, we saw more as better, so the data stockpiling combined with cheaper storage combined to create massive data stores. PC's came on the scene and continued to explode through barrier after barrier: faster, cheaper, and "ubiquitous". The talk was of the number of PC's per worker, and the forecasts of deployment to the desk of every knowledge worker appear to be very accurate. The 80's were also a time of limited use of graphics in the everyday operation. Graphics were the things of special occasions - annual reports, etc. CURRENT MANAGEMENT INFORMATION ENVIRONMENT Now in the 90's we are often drowning in data. How do we turn this wealth of data into information? What is the cost of this transformation? Bill Inmon, regarded by many as the father of the data warehouse, has noted that the unresolved issue is the transformation of distributed data into usable information regardless of the platform.(Knight, 1991) The business areas at our institutions have acquired extensive technical capability. The staff are increasingly computer knowledgeable, especially using desktop computers. Many Information Systems organizations have established the policy that the business areas are responsible for producing their own reports, a step which the University of Massachusetts President's Office took in 1992. Regulators are also stocking their databases and expanding their reporting capabilities. Explaining findings derived from our "exported" data taxes our institutions ability to respond rapidly with in depth and comprehensive reporting and analysis. According to the National Center for Education Statistics, "Various legislative requirements are also leading to the development of new data sets for higher education.[such as] the 1992 Higher Education Re authorization Act, as well as the Student Right to Know Act of 1990 which give new importance to the need for accurate and comprehensive information." (SHEEO/NCES,1993) Business graphics have become essential to credible communication, again, the result of years of desktop computer deployment, friendlier packages, more skilled users, and everyday use in newspapers, newscasts, business weeklies and even political debate. Our current environment is one of demand for rapid information delivery. Increasing experience with the responsiveness of desktop computers has led the business user to view timely delivery of information as an entitlement that should be within the abilities of their information suppliers. If we are to meet the information needs of our administrators, we must act now, leveraging our resources to deliver more capabilities in the short term, while building toward a service environment that keeps pace with the demand for management information. LEVERAGING STRENGTHS At the University of Massachusetts we began by assessing our core competencies as they related to delivering information access and reporting in a client/server approach. That is, an approach that provides requested data from a designated source computer (server) to desktop workstations (client) applications. In our environment this was assumed to be a VAX server accessed by desktop computer application software. The areas of strength that we identified as essential to success were infrastructure, data, business area readiness, and management support. Our infrastructure strengths were in the area of networking, a critical factor for our dispersed environment of six sites throughout Massachusetts. In 1993 we implemented a wide area network using a T1 ring. The networking services group has extensive experience in multiple protocols on the wide area network as well as experience in local area networks consisting of both Apple and IBM compatible desktop computers. Experience in configuring workstations, supporting desktop computer software, and administering a local area network for several years were also recognized strengths. The university had made an effort to improve the management of the data resource through the efforts of a university-wide Information Systems Task Force, which adopted a resolution that data administration should be a priority. As a result, a University Data Administrator was named in 1992. The official focus on data as a key resource led to improvements in the data dictionaries for the primary systems administered by the President's office. Particular attention was devoted to the Human Resource Management Information System(HRMIS), which provides all payroll and on-line human resource processing for the university. In addition to supporting this processing with operational reports, HRMIS also provides detailed data to the Commonwealth of Massachusetts. In 1992, University Information Systems senior management decided to form the University Information Resources organization. This organization was charged with promoting University-wide data administration and implementing a data warehouse to improve management reporting capabilities. The primary business area supported by the University Information Resources is the President's Office. Secondary support is provided to the campus administration through the Vice Chancellors of Administration and Finance, Institutional Research Office, Budget Directors and Human Resource Administrators. Notably, these constituencies were making extensive use of desktop computing and, in many cases, batch ad hoc report generators accessing legacy system extracts. While there was a history of commitment to management reporting and the development of a data warehouse, the work prior to December 1992 was limited to research and fact finding. The activities from December 1992 through the Fall of 1993 included implementation of the data warehouse, client/server reporting, ad- hoc reporting and an Executive Information System. The strategies and lessons learned may offer ideas and examples of how success can be achieved in developing and implementing client/server reporting to promote effective administration. A SOLUTION IN THE REAL WORLD Given the pressing business demands, it was clear that a protracted development process was not a viable approach. While there was preliminary support for the concepts of data warehouse, data administration and management reporting, it was clear that continued funding would require visible tangible results. Certainly the core of this client/server project was the business need for management information. The strategy employed was to identify a single area of business need, and focus the project on implementing a generic solution. It was important that solving the particular business problem require us to address the essential components of client/server access to a data warehouse. One of the early advantages realized through our focused approach was positive movement toward a clear goal. I contrast this to a protracted planning process that might require an early consensus that is extremely time consuming. An example of a task that could delay progress is deciding the full range of data to be included in a data warehouse. The answer to political issues such as this could easily require months, if not years, of meetings with representatives from many functional areas. Our strategy is to develop the data warehouse in stages, consistent with an evolving University Data Model. One problem area that came to light in early meetings with key business managers was that of human resource information. Some standardized monthly reporting had been developed to provide key measures and indicators for comparison across the five campuses. However, the reports only provided information for the current month. Over a period of three years, a spreadsheet application grew around this monthly data. The application, which was developed in the President's Office, grew to a set of eight reports that accessed an Excel database to calculate comparisons to prior periods, average salaries, and permit trend graphing. The data entry from the standard production report, and the manipulation of massive spreadsheets made this a high cost application which was identified in several discussions as needing a "better approach". Since the business area staff had clearly established Excel as their standard, we decided that the initial client/server implementation should be able to deliver data to Excel spreadsheets with minimal user effort. The decision to implement a client/server solution was made basically from the requirement to deliver a solution that offered easy access to critical data, and recognized the need to provide for more extensive reporting capabilities. Some of the clearly identified reporting needs were in the area of more detailed Human Resource data as well as financial and academic data that knowledgeable business staff could combine as needed without specialized technical skills. The need to provide this type of cross functional reporting clearly indicated a relational database would be required. While the majority of our legacy systems reside on an IBM mainframe, the IBM environment does not have a relational database environment. However, our VAX environment had Rdb as an installed supported relational product. During some of the early research into improved management reporting strategies, Digital Equipment Corporation had provided extensive information describing their implementation and use of Rdb for massive data warehouses. As previously noted our networking capabilities and support were very good, especially in the options available through our VAX environment. We decided that the warehouse would use the exact data that was reported in the standardized monthly reporting. This would provide more detail than was available in the spreadsheet database, and would allow us to start the warehouse with over 30 months of history. The general information needs identified were: ad hoc access to produce summary reports and comparisons, the need to provide executive information, with perhaps some direct hands-on executive access, and an easy way to move data directly into Excel spreadsheets for manipulation and graphing. Our early research had identified that Data Access Language(DAL) was a reliable widely supported facility to use on the VAX to enable client machines to access data stores on a VAX platform. Testing indicated that indeed Excel and DAL worked quite well together, so the combination was identified as the first implementation of client/server reporting from the data warehouse. The decision to utilize Data Access Language(DAL) precipitated a number of other decisions. One notable area of some uncertainty was software for ad-hoc reporting. Years of experience in non- workstation based access tools indicated that better service to our customers would require a client/server solution. Once DAL was identified as a standard client/server technology, several DAL compliant products were identified that would provide highly flexible ad-hoc reporting. In summary this project answered a critical data need expressed by some of the top administrators in the institution, by providing easy access to well defined data using familiar tools. This has proven to be as successful an implementation as this summary would suggest it should be. A variety of other factors contributed to the success of the project. The remainder of this paper describes some specific lessons learned and provides guidelines for other key success factors. LESSONS LEARNED 1. Once the decision-making process began, a clear framework emerged, with of course certain limitations . However, initial decisions such as selecting the database management system, focused the thinking of the team on implementation tasks and time frames. It also become much easier to identify activities that were not essential to the initial implementation. 2. Client/server need not be viewed as a radical change of direction, but rather an extension of other initiatives. We had clearly decided to implement a data warehouse, but had not established that the initial access would be client/server. 3. It is not practical to become expert at everything that is listed in the technical critical success factors for client/server. You may need to develop alternatives such as an interim solution, or a partnership with an expert. One of the key components of our data warehouse is an encyclopedia. Our design called for the encyclopedia to be integrated into the desktop environment, such that business rules and definitions could be viewed, and queries could be generated via navigating the encyclopedia; Aetna Insurance has developed a facility that is highly successful. However, this functionality could have added as much as 6 months to the implementation, so we developed some standard queries that could be invoked to provide the core information. Since our Rdb expertise was lacking, we partnered with DEC who provided some long term design assistance, as well as some rapid delivery of Rdb services. 4. The specialized knowledge, the need to have innovative approaches, the need to build ongoing support in the IS group requires a cross discipline team. We included the University Data Administrator, a networking and architecture specialist, and an end user computing expert as part of the core team. This worked very well in raising the concerns necessary, and in building commitment in these essential areas. 5. Partnerships with user constituencies are invaluable. We plugged into some key university-wide groups that would be stakeholders in the data warehouse, such as Institutional Research and Human Resource Administrators. The number one concern of these groups was ensuring proper definition and understanding of the data that would comprise the data warehouse. Groups such as these offer help in definition work, as well as provide essential feedback. One way of gaining support for data warehouse, or similar activities that make data accessible to management, is to make the implementation a "win" for these constituencies. Whatever capabilities we provide for the President's Office, we also provide to the campus. Although the campus may already have the data, the access via a new client/server tool may be welcomed. For example campus access to President's office EIS reports and graphs, provides these facilities without any development cost, limited of course to the data that the campus is authorized to access. 6. You do need to pay some attention to the "open systems" approach. Open solutions can provide flexibility and lower implementation costs. Open architecture, open systems is really driven by what your environment is, and what you expect it to become. Why wouldn't you strive to have an open solution? Do I really want to develop one application for Macintosh and another for DOS platforms? These are some of the issues which client/server forced us to answer. We have decided to require portable applications. This is working out quite well using open solutions such as Excel's database interface, and the HOLOS Executive Information System. You may also find that by starting into client/server, you will actually act as a catalyst for the organization to set a direction toward open architecture. CHOOSING INITIAL APPLICATIONS - THE BUSINESS CASE The popular phrasing applied to newer approaches seems to be, "it's more of an art than a science". However, there are some basic approaches that can make choosing an initial client/server application more science than art. No doubt as your reputation, time and money are going to be invested in the first client/server application, it would be nice to find a "sure thing". The easiest aspects to address is what types of applications should be avoided. If client/server is new to your organization, there are a number of areas that must combine to make the effort successful. Although client/server has some similarities to what you might have done in the past, it is deceptively different. It is more than installing a new mainframe reporting package, it is more than installing a new PC package, or a new network access, and yet it often includes all of these activities. For this reason, it is recommended that the initial application should not be: mission critical, date driven, or a large application. A somewhat more conservative approach is to "concentrate on a class of applications that do not require real-time data access between the microcomputer and the mainframe because these applications are more easily manageable, fit better with the current skill level of most organizations, and can be delivered more quickly." (Capraro, 1993) Aside from the inherent risks in these projects, cost can be a significant issue. Frank Dodge, co-founder of McCormack & Dodge Corporation, in responding to a question on cost savings realized with client/server solutions, indicated that "it's a mistake for companies to believe that overall costs they allocate to a particular set of applications will dramatically decrease within the first eighteen months or so." (Dodge, 1993) While the technical components are clearly important, the best technical development of a low value application will not produce a successful result. One of the best approaches to finding a high value application is to be engaged with the business areas you support. Involvement in the business issues and understanding the demands being made of the business areas, provide a basis for answering the central question "Where can I provide a new information service that gives the business area a boost in productivity or information access? " Why consider productivity first? The stories are legion about the reams of paper that are printed from legacy systems, distributed to the business areas, and then used as a source for a few key data entered into a desktop computer. The desktop computer can then be used to produce the report that has real value to management. I would suggest that this is perhaps one of the top candidates for an initial client/server application. Looking at some key success factors shows why a focus on this type of data re-entry situation can offer a high probability of a successful client/server solution. Business Need - With resources generally limited, the reporting topic is likely to have some importance if it is being maintained in spite of the cost and delay of data entry. The topic area offers opportunity to improve quality, timeliness and the scope of the data available for analysis and accuracy. Well Defined Data - Whenever business area access is provided to a data source, the data should be clearly defined. Client/server can accentuate the need since the particular solution may require distributing the data to local servers. Data that is reflected in standard reports is often defined in the "business rule" or logic that produced the report. In addition, data that has been available for a period of time in a standard report usually has an accepted definition. The availability of these types of de-facto definitions enables the client/server project to proceed without needing to invest immediately in potentially lengthy data definition sub-projects. Easily Accessible Data - The fact that report programs exist provides an easy opportunity to extract, at a minimum, the same data that is included in the standard reports. Data distribution and access policies can be less of an impediment if the initial client/server application is simply deploying new capabilities to areas that already have paper access to the data. Historical Data - While typical operational system reports focus on the management of the business function, operational reports do not provide the type of trend information needed for decision support and executive management. Often, these management reports require a business area to key enter data to create local private "databases". A client/server solution can address this need, by providing a much easier means of drawing on the data, especially if the solution is to convert the data to a relational data store. There are two areas of human factors to consider in choosing the initial application - the business partnership, and the staff expertise. Business Partnership - If management in the targeted business area is an ally in the project, then the likelihood of success is increased. While building on past successes is a definite advantage, strong allies can be established by developing a shared view of the benefits. One way of enhancing the benefits is to have as little disruptive impact on the business area as possible. The single most important area to address in minimizing the impact is to avoid solutions that require major changes to the business area, such as conversion from PC to Macintosh, or from one spreadsheet package to another. Staff Expertise - Client/server is an evolutionary step in information service delivery, particularly for reporting applications. This means that there are similarities to many of the services already being provided. Staff who have been heavily involved in building databases, supporting ad hoc data access via 4 GL tools, supporting desktop tool installation, and system design are key resources for a client/server application. In addition, skills are required in networking and desktop network access, which may necessitate expanding the team beyond the system development group. At the University of Massachusetts, the data warehouse team, responsible for implementing client/server reporting applications included key infrastructure staff. These infrastructure experts addressed the networking, desktop connectivity and many open architecture design issues. Partnerships with vendors - Vendors can bring very specific expertise to the project, which allows the team to focus on strengths, rather than trailblazing in new technical capabilities. One example of vendor support which was highly effective in the University of Massachusetts' client/server implementation was a one day project review with a Digital Equipment Corporation Team. The DEC team included experts in relational database design, wide area networking and desktop connectivity. The review provided some expert advice, and an extra measure of confidence in our project plan. We later contracted with DEC to develop an EIS client/server application using the HOLOS software developed by Holistic Systems Corporation. While the University of Massachusetts team was responsible for all other components of the EIS project, it was extremely helpful to use DEC's expertise to develop our first HOLOS-based application. COST SAVINGS AND JUSTIFICATION The debate of whether or not client/server and downsizing save money will probably continue for some time into the future. However, in focusing on reporting applications, we can rely on some clear business issues to determine if client/server is a cost-effective solution for a particular information need. We can consider some of the common applications and issues to illustrate how cost and value might be presented to support a decision to implement a client/server solution. A key question to answer is "how is critical reporting being addressed?" If data is being key-entered to desktop computers to create reporting databases, there are some associated costs. In addition to the data entry cost, there are some quality control costs in checking for keying errors. If the keying errors are not detected, what is the cost of decisions based on incorrect data? While this cost cannot be quantified as easily as key entry time, raising the question may be a powerful factor. Data entry report applications can be problematic due to the inherent delay in producing the report, and the potential loss of some important information that was deemed "too much" data to be keyed. These situations lead to decisions being made without the benefit of a key resource, the institutions' data. There is also a high potential for duplicate effort since many recipients of a valuable standard report may use the report for similar purposes such as tracking the same trends via their desktop computers. The nature of management reporting applications often implies use of central data stores with distributed access, a standard model. When this model is not used there is increased risk not only of duplicate effort, but of arriving at different answers to the same question, depending on the design, quality, and currency of the redundant, locally created data sources. Even given consistency, if complex rules are needed to evaluate or manipulate the data, these rules need to be replicated for each locally created data source, further increasing the risk of erroneous results. A key characteristic of client/server applications is the familiar and comfortable desktop access they provide. Training business area staff to access data from graphical interfaces and desktop query tools with context-sensitive help is much simpler and faster than delivering training and support for 4 GL languages. The cost savings for the business area is in both the reduced training time and the more effective use of the reporting capability; the difficulty and time spent making 4 GL's produce desired reports and graphs is dramatically reduced in client/server approaches. While there are a variety of ways that cost savings could be approached, improved information quality and effective use of scarce resources, particularly in the business areas, can be a powerful illustration of the benefits that can be derived from a client/server approach to management reporting. IMPLEMENTATION STRATEGIES Creating a Focus Once you have decided to proceed with a client/server solution to a key information reporting need, it is essential to create a focus that will assure success. Although the justification for the project may be that a number of business areas will benefit, the initial implementation should be limited to a small number of customers in the same location. Allowing for a few problems is realistic and quite manageable if you have targeted the customers who will succeed with you. Undertaking the initial client/server implementation with a customer who does not share your belief in the project, is inviting intolerance and negative publicity that can brand a project or technology as a failure. One way to maintain the targeted area's support for you and the project, even if you are experiencing some delays or problems, is to be highly visible in the customer's area. Frequent demonstrations of prototypes, validation of workstations, networking tests, etc. all show that the project has commitment, and invite a supportive reaction from the customer, even if the project is off schedule. Again, it should be emphasized that the potential for problems is one of the reasons that the initial client/ server application should not be "mission critical". Client/server applications are well suited to prototyping and therefore promote satisfied customers, provided that the developers pay close attention to customer reaction to issues such as ease of use and features the customer deems critical. Marketing The initial client/server application represents an important event in both the business and Information Systems organization. A successful implementation will require support from many individuals in these organizations. Since people can not support what they don't know about, it is important to keep these constituencies informed. Using a variety of communication opportunities, as often as these opportunities are available, will help to ensure that everyone is aware of your client/server project. Updates in newsletters, staff meetings, departmental meetings, and informal updates are all effective in ensuring that the project is visible, and a recognized priority. An important aspect of the communication, particularly for the business areas, is to set realistic expectations, and continue to reinforce these expectations regularly. This is especially important if the implementation will occur at only one of several planned sites well in advance of the implementation at the other planned sites. In this situation, and as a general means of sharing information, demonstrations of the client/server application are a powerful communication tool. Since an amount of skepticism is often attached to new initiatives, demonstrations serve to dispel the notion that the whole effort may be vapor- ware. Demonstrations also help to clarify issues and surface key concerns before the project is delivered. For this reason, demonstrations within the Information Systems area may be conducted very early in the project and may show only product features, since application and development may be in very early stages. Once the project has reached final testing stages, demonstrations take on a different role. At this point they can serve as a "wake up call" to everyone who will be involved in the implementation. This can help to create a push to wrap up any outstanding tasks. The demonstrations of production-ready systems also create enthusiasm in the business areas, which is important to generating more interest in future applications. Infrastructure The network plays a key role in client/server applications since the client workstations must be able to communicate precisely with servers, whether on local area networks, wide area networks, or a combination. For this reason, the infrastructure support is more critical to success than ever before. New areas of support that should be considered with networking include workstation configuration and workstation product installation, particularly for products that support the network communication. Whenever possible, the infrastructure support areas should be offered the opportunity to participate in product selection. Involving these areas at the inception of the project and ensuring a continuous update process will pay dividends in a client/server project. SUMMARY While client/server development can appear daunting at first, there are strategies that can be used to introduce this technology with limited risk and expense. As with any new technology, success should be measured by the value that the application of the technology brings to the institution rather than the inherent potential of the technology. This paper has attempted to provide some insights into one approach that was highly successful in delivering business value to the University of Massachusetts. References Capraro, Anthony J. "So You Want To Downsize Your Applications?" Information Strategy, Spring 1993 pg. 54 Dodge, Frank "Client/Server Technology: Looking Beyond the Mainframe." Global Business Review, Digital Equipment Corporation, 1993 pg. 25. Knight, Robert. "DBMS - Facing '90s Obstacles." Software Magazine, v .11, December 1991, pps. 27-28. SHEEO/NCES Communication. "Network News." v.12, no. 3, September, 1993, pp. 1-9.