Far from being more than a repository, IBM’s AD/Cycle is rather less Le Repository est arrive, but most people are unsure what to make of it. One of the first reasonably considered responses was a seminar held in London by the independent consultant Martin Healey. Planned several months in advance as a preview of what […]
Far from being more than a repository, IBM’s AD/Cycle is rather less
Le Repository est arrive, but most people are unsure what to make of it. One of the first reasonably considered responses was a seminar held in London by the independent consultant Martin Healey. Planned several months in advance as a preview of what The Repository Announcement was likely to contain, Healey’s papers were hijacked by events. Nevertheless, aided and abetted by representatives of third party software vendors, Healey had an analytical stab at the implications of AD/Cycle. He started by turning IBM’s declaration on its head, suggesting that instead of AD/Cycle being more than a repository, as IBM claims, it was less than a repository, since it was a computer-aided software engineering environment only with nothing more than a statement of direction from IBM as regards networking (CI No 1,268). Healey believes it took so long for IBM to make an announcement because it had tried to design a totally consistent information base but was having problems making it fit with run-time network management and so came out, initially, with a subset of requirements for the software engineering environment. Other IBM products will come out and be integrated in time, but over the next two years the number of management functions that need to be integrated will increase.
AD/Cycle is pitched at senior managers
Healey asserts that the main thing to bear in mind when evaluating the AD/Cycle announcement is that it is targeted at senior management. It is intended as a business solution rather than as a product push. For example DevelopMate is the first repository tool to be addressed by IBM and it offers a co-operative interface between the businessman and the system analyst. It will doubtlessly be marketed as a way to knit corporate management together.
Despite all the partners flocking to sign, AD/Cycle may prove bad news for small firms
While the announcement of AD/Cycle may leave companies like Oracle Corp and Micro Focus Plc relatively unmoved at the moment, it has caused hysterical excitement among smaller third party software houses, which almost all claim it to be an unmitigated benefit to their business. In fact they may live to eat their words. At first sight, for example, an acquisition driven company such as Computer Associates International Inc may welcome a standard platform since it holds the promise of a clarified sales pitch of an externally imposed integration of software engineering products. In fact, however, as experienced programmers know, tools that have not been designed from scratch to work together do not work together very well. There are other reasons why members of IBM’s software co-operation programme may not fare as well as they think. For although the AD/Cycle boasts an open architecture (as yet unpublished) it is first and foremost a DB2.2-based product and will therefore favour IBM in the long term. Furthermore, the companies that have rushed out with interfaced products (virtually all 30-odd companies included in the enablement programme) rather than waiting to develop fully integrated DB2.2 products may fall prey to more competitive products from companies such as Micro Focus particularly as nobody will have AD/Cycle up and running till the end of 1990 at the earliest. That being said, certain companies described by IBM as marketing partners got a lot of publicity at the time of the announcement. According to Keith Short of James Martin Associates, all companies on the enablement programme (to which JMA belongs) will get the same advance information as KnowledgeWare Inc and Bachman Information Systems Inc. As marketing partners on the Co-operative Software Programme, however, KnowledgeWare’s Information Engineering Workbench tools and Bachman’s Database Administrator tools will be marketed by IBM itself. Yet Short was confident that JMA’s position through the marketing of Texas Instruments’ IEF Information Engineering Facility family had been endorsed by the AD/Cycle announcement. The IEF toolset is, after all, an in
tegrated set that develops applications for any run-time environment, rather than, say, that of Oracle which builds applications for Oracle’s own run-time environment. Consequently, JMA can with some justification claim a backward compatibility with the AD/Cycle as well as an advantage over companies without an integrated toolset. Meanwhile, Mike Gilpin of Learmonth & Burchett Management Systems Plc argued that LBMS Information Manager was a compatible superset of the AD/Cycle, offering full cycle support including implementation links to fourth generation languages such as Pansophic’s Telon and IBM’s CSP. He is of the opinion that the repository will eventually be offered under OS/2 although the IBM announcement implied that it would remain remote on the mainframe or minicomputer. A Mercer of Pansophic thought that the OS/2 implementation of the repository must arrive quickly because of the growth of the personal computer in application development. Pansophic, of course, has its own integrated toolset that runs under OS/2, and the company was considering launching its ogn repository last year.
How does IBM plan to quarantine Unix?
As part of the five-layer AD/Cycle, IBM has made a declaration of intent to include knowledge-based systems in layer three. Healey said that he was aware that IBM was working with other unnamed companies to develop pro forma or skeletal designs for applications that will be kept in the repository. This would lead to more stable platforms, reduce design efforts, and has to be considered a step in the right direction at a time when software re-use is still off the cuff. Something that intrigued Healey was how IBM would police programmers moving such designs, written in Cobol, across to Unix via the relevant compiler…? Healey also thought it significant that IBM had announced CSP/370 Runtime, arguing that this was the first time IBM had attempted a code generator, and that following this statement CSP could well become the standard fourth generation language.
A rose by any other name…
Finally, a rider to initial responses to the AD/Cycle: nobody at the seminar believed IBM spent a great deal of time dreaming up the brand name, whatever IBM says about the matter. After all, what company really wants to submerge a new product in toilet humour? But there we are, the repository is already known as the suppository – AD/Cycle only adds to the problem. – Katy Ring