IS&C Strategy for Selecting the Hardware/Software Environment for New Applications:

In responding to the customer survey conducted by IS&C during the summer of 2001, several people asked for more details regarding our strategy of building new administrative software applications both "on" and "around" the mainframe. An expansion follows.

Simply stated, our strategy is to use the mainframe environment when it makes most sense to use it and to use alternative technology when that makes the most sense. "Sense" is determined by the analysis that we do with our customers at the beginning of any significant project and includes such things as cost/benefit, extensibility, ease of use, and compatibility of technologies. Over the past ten years IS&C has diversified its technical platform offerings such that we now support UNIX and Windows 2000 in addition to our OS/390 mainframe services. Currently we support more than thirty separate computing platforms running dozens of applications in the UNIX and Windows 2000 environments, so we have many alternatives to consider in our analysis.

Said another way, we embrace a modified version of "best of breed". When faced with a new application opportunity, we work with the functional office to look at all of the options and recommend buying or building the best system possible considering our own unique history and the available resources. Given that few applications will stand alone in the emerging administrative software milieu, we must choose an application environment that permits integration with what already exists at UCSB and, as best we can, what might be coming next. This imposes few limits on our selection, given that we already operate all of the major operating system environments currently available (Windows 2000, Unix/Linux and OS/390).

In our view, the emerging software environment is characterized by multiple tiers or layers, each containing one or more components that are separate, but which facilitate integration into a larger whole. As each functional office attempts to automate more of its business processes while at the same time meeting more of the unique needs of each of its individual customers, the complexity of the software supporting its business processes grows. It is impossible for a small software group to create all of the software that is needed by all of the functional units at UCSB. Consequently, we must buy some components and we must build some components and we must integrate them all into a working whole that supports the combined business of the campus. Focusing on components will allow end users considerable freedom of choice in acquiring software and will minimize the impacts of change over time.

Implied in this strategy is the proposition that mainframes are not nearing the end of their useful life and, therefore, should be considered as a potential platform for future applications. We in IS&C believe this is true. In our experience, the mainframe continues to provide unmatched levels of reliability, security and availability while performing as an efficient host for multiple enterprise software systems. While there have been many technical innovations introduced to the marketplace since the introduction of contemporary mainframes in the sixties, none has completely supplanted the mainframe for general purpose computing and for the support of large-scale transaction systems. Indeed, the mainframe has continued to evolve, becoming more flexible and more cost-effective as time has passed.

On another front, we also believe that the internal structure of our mainframe database (Adabas) is much better suited to supporting new database technology than other current products. By its nature, the hierarchical, highly indexed design of Adabas makes it a good object repository, supporting current object-oriented software environments better than relational database alternatives. Thus, Adabas provides an excellent "bottom tier" in the n-tier, component structure outlined above. Software AG, the maker of Adabas, has capitalized on this turn of events and has also captured the lead in XML development with their Tamino product. (XML promises to be the "Esperanto" of database systems, providing the common language by which computers can exchange business data while performing joint transactions.) Moreover, the recently released version of Com-plete, the Software AG teleprocessing monitor, enables direct web screen (html) interaction with Natural programs running under its control. Consequently, the future of Adabas looks even better now than it has in past years and our mainframe Software AG environment might serve the campus well for decades into the future.

One major issue surrounding the mainframe has been the perception that it is more costly than other solutions. Our cost data based on the operation of a variety of systems over the past several years indicate, in contrast, that the mainframe can be the most efficient solution available when appropriately used. Appropriate uses are determined during our pre-implementation analysis process. Consequently, we think that including a mainframe in the mix of platforms we support is a good strategy for the indefinite future. This strategy allows us to take advantage of economies of scale in staff support and to continue to run legacy applications for as long as they are useful to the campus while still providing opportunities to match each new task to whatever platform alternative is most desirable.

Therefore, we plan to build "on" the mainframe when significant portions of an application already exist there or when we need a high-performance, back-end data store. We plan to build "around" the mainframe when the desired software runs more effectively in an alternative environment. Examples of the latter tactic include CorporateTime on a Sun platform and the implementation of the Data Warehouse on AIX and Sybase. In the Data Warehouse a perceived weakness of Adabas (the difficulty of extracting data from its highly indexed and hierarchical structure) is ameliorated by the more easily accessed Brio/Sybase system running on Windows 2000 and AIX. This exemplifies our strategy of building "on" and "around" the mainframe.

Building "around" the mainframe also implies that we can connect newer systems to our older systems without extraordinary effort. To accomplish this we have installed another Software AG product (Entire X) that makes connections among systems much easier than ever before. With Entire X we can implement applications on remote platforms that seamlessly access and update data in Adabas, thus gaining the best of both worlds. The new Travel, BARC and DARS applications all use Entire X (or its predecessor, Net-work) to access Adabas. Further, Entire X is at the bottom of an authorization scheme that allows web programmers around campus to make certain that the users of their systems are authorized to perform secure functions based on data stored in Adabas on the mainframe. These examples truly exemplify our strategy of building "on" and "around" the mainframe - even within the same applications.

In summary, we in IS&C believe that we offer a great mix of platforms, technologies and services and that we are well positioned to support and integrate whatever administrative software applications the campus might acquire or develop in the future.

Glenn Davis
Director, IS&C

For assistance or further information please contact webcontact@ucsbuxa.ucsb.edu .

Last Modified: CGH,12/28/01