Since Computer Associates Inc became the biggest software company in the world, everyone has been wondering what it plans to do next. And while the industry marvels at CA’s ability to absorb wildly varying software products and keep them on track (at least financially), guessing how CA goes about it becomes a favourite passtime. But being one of the more timid giants in our industry, CA rarely comes out in public and lets anyone know what’s on its mind. It doesn’t break its revenue out across mainframes, minis and personal computers, and most still think of it as a mainframe only software house. All the more reason to sit up when a dignitary from the US visits an overseas outpost like London.

Full frontal

Darrell Olps, in charge of development at CA’s Micro Products Division came over mostly to talk about IBM’s Systems Application Architecture. CA has decided on a full frontal approach to SAA, pursuing it across its entire micro product line and expecting to introduce these products over the next few years, promising SuperProject first, and SuperCalc and others to follow. Olps has some interesting observations on SAA. Yes, SAA is very difficult and costly to interface, and No, I wouldn’t have chosen to do something so ambitious if we had not been forced to by IBM’s introduction of SAA, and We couldn’t get away with just rewriting the interface on SuperProject, we had to change the whole concept of the program from interrupt-driven to event-driven, as well as adding all the necessary graphics. Olps is probably one of the best technicians to put onto any graphics issue, having been inherited from ISSCo, itself a leader in the graphics field, when CA took it over in 1986. SAA on the whole makes life more difficult; it gives me some tk/6s that I wouldn’t have, but it makes some decisions for me that I would rather make for myself. On the subject of continuity in his micro products interfaces Olps is confusing. He says that by complying with SAA, SuperCalc will end up with an identical look and feel to Lotus 1-2-3 and other SAA-compliant spreadsheets. But he insists that the function of the product will remain unchanged despite a completely new look and feel. Inevitably CA says that it will differentiate its products by this superior function under the identical format, and by offering exactly the same product under other environments, such as DECwindows on the VAX, X Window on Unix and under straightforward MS-DOS. But no more than a long hard look at Unix was promised for CA’s products, and Olps gave the impression that no work was under way in that arena as yet, hinting at a company looking for standards to follow in a standardless world, citing 13 different varieties of Unix and none yet worth following, despite CA being a confirmed member of the Open Software Foundation. Perhaps the effort of following so many different types of interface will tell on CA’s portfolio of products. It seems obsessed with spreading them across the widest possible number of environments, rather than adding function to existing products, or adding new ones. (When was the last time CA wrote an original product?)

Although Olps wouldn’t be drawn into discussions of how CA manages to keep hauling in alien software and extracting growth out of it, reading between the lines with the help of the UK management, more eager than in past to tell its story, a picture of the CA corporate culture and its long-term objectives become visible. In the UK 55% of CA’s UKP46.8m turnover came from mainframe software sales last year, the rest coming from mini and micro products. This is down from 70% two years ago and shows that CA isn’t quite the mainframe specialist that it used to be. The last of those two years CA was the owner of ADR, and included some of its predominently mainframe sales. If this ratio were true throughout the world, it means that the mini and micro software business is doing close on $450m. So does that mean that the mainframe packages are dead or dying? CA has ADR’s development tools and its Datacom/DB database, the Cullinet I

DMS and IDMS/R database product lines, Uccel’s data centre management products, and even Olps’ native ISSCo mainframe graphics software, not to mention its own sort programs, its job schedulers and job stream submission products, its security and optimisers for which it is traditionally known – not to mention its own CA Universe relational database. The answer seems to be that there are so many products with a CA label to help run a mainframe centre that they are beginning to be grouped together in discounted inter-related bunches and sold en masse. One concept it plans to unveil over the coming weeks is what it hopes is a quantum leap forward in operations management. A reliable, integrated system for running the operations at a mainframe site, which CA refers to as the lights out approach. (No people needed.) After all, said one CA executive, you only want your operations people to be involved in serious problems. Software should be able to run the department for most of the time. And why have that person present? He can be at the end of a paging device with his laptop handy, and only do any work if a problem comes up the software can’t handle. Naturally, CA doesn’t want to address directly the numbers of DP staff this would leave in the dole queue, but it is certain to feature in the firm’s benefits breakdown once the salesman gets his hands on the idea. What of the three relational databases, Datacom/DB, CA-Universe and IDMS/R? This week it plans to issue a 73 page document to its users, explaining that CA plans to keep IDMS/R development separate from the other two, but that Universe will gradually merge with Datacom.

Big Bang

CA thinks there are a few functions in Universe that need to be in the next release of Datacom, and instead of its usual conservative approach to evolving its products, this merger will be more of a Big Bang, occurring in a single new release delivered late in 1990 – and more development effort for the Universe users than the ex-ADR customers looks likely. The next stage should see IDMS and the new merged product built up to a similar functional level, although none of the changes would mean a conversion to the other database. CA also expects great things from the tools that surround Datacom and IDMS for generating DB2 and Rdb applications. It seems likely that IDMS/R and its tools will become its DEC and perhaps its Unix offerings, while the Datacom/Universe product would be the lead product on IBM mainframes. CA is nervous of any speculation like this, because users of IDMS/R on IBM mainframes may ask whether their product could be expected to keep up with market developments, and with reason may expect to see it left behind in terms of innovation, so that they hav no choice but to move to DB2 or Datacom software, or even down to DEC hardware. Questions like Does CA have any real net growth? Or is it all growth by acquisition? and What happens when there is no-one big enough left to acquire, will remain unanswered by CA staff, but any right-minded guru must look ahead to that day and worry if there is a strategy to take it past $2,000m.