It is 12 months since IBM described the annual show at Birmingham’s NEC as the largest user exhibition in the world. But recessions mean that even the mighty are tightening their belts, and Tony Cleaver now acknowledges that IBM ’91 is merely the company’s biggest showcase in the UK. New-found modesty aside, IBM ’91 was remarkably similar to IBM ’90. Certain traditions were honoured, and the whole shebang was opened by an underemployed politician convinced that training is the key to a prosperous future. Sir Norman Fowler was introduced as the local Member of Parliament who, despite his retirement from Cabinet unlike Old Nick, he wanted to spend more time with his family has a keen interest in training and technology. Just in case anyone doubts that Sir Norman still has friends in high places, he forecasts that inflation will fall to around 5% by the year-end, interest rates will come down step-by-step, and the second half will see a return to growth. It’s worth recalling that when Lord Strathclyde opened IBM ’90 he claimed that the UK faces three principal challenges. The first was international competition from Japan, the Pacific Basin and Europe.

Sir Norman

The remaining challenges were demographic changes and technological opportunities, and it’s technology which offers flexible approaches to training that will create a highly skilled workforce for tommorrow’s world (CI No 1,376). Sir Norman says that irrespective of the economy in the next six to 12 months, there should be no doubt about the three principal challenges facing the UK, and what with international competition, demographic changes, technology and training… well, you already know the rest of his speech. With that familiar litany, IBM ’91 was declared open and according to Tony Cleaver, it was no less than a reflection of the new IBM. Despite grumblings about enough hot air to float a balloon, one genuinely new feature was the plethora of seminars, over 300 in all, and quality varied from the informative to nothing more than a sales pitch. Deborah Thomas, one of IBM’s AIXers, addressed openness and standards, and she says that the problem with open systems is defining what they actually are, and that definition may depend on which particular trade magazine lands on your desk. The beauty of IBM’s open strategy, claims Ms Thomas, is that it embraces both Unix and non-Unix worlds. Any movement towards open systems has to take account of current investments, and the IBM premise is protection with added functions. The recipe is Systems Application Architecture, plus multivendor enablement, plus IBM’s Open Systems architecture. Systems Application Architecture is key to IBM’s open systems direction, and the three underlying features are interoperability with non-IBM systems; support for industry and de facto standards; and SAA-availablity of many interfaces that are currently available under Unix. Interoperability between SAA and AIX remains a vexing issue, and despite the growing number of applications that run under both, IBM will insist that the process is a bridging one. –

By Janice McGinn

There are common functions like Fortran, C, and Cobol, and there are bridged functions like NetView, while additional Unix options include TCP/IP, X Window and Sun’s Network File System. Deborah Thomas attempted to outline IBM’s future directions, and in the area of multivendor networking, says that the company intends to support different industry standard protocols, and to build an application interface that’s independent of the protocol used. When it comes to data access, she claims that IBM will campaign for its Distributed Relational Database Architecture, published in July, but if that is not generally accepted, then IBM will conform to an industry standard. On remote application access, the IBM direction is OSF/Motif and an OSF/Distributed Computing Environment for both AIX and SAA. A consistent user interface, an absolute must for office systems although no one has emerged as yet, also features in IBM’s directions. Deborah Thomas talks of an an architecture

that will facilitate Common User Access and Motif convergence, which means the convergence of an SAA pillar with the Open Software Foundation interface to AIX/Unix. These various directions give the impression that SAA and AIX are about to converge, despite IBM’s denials in the past, but MsThomas refuses to say that the RS/6000 will ever be part of SAA. However, John Glyde, AIX Business Manager, is more frank. No, he won’t say that AIX is to come under the SAA umbrella, but there will be a gradual coalescence, which generally means growing into the other, or a fusion of sorts. That’s different from what IBM was saying in July when Mike Saranga, assistant general manager of IBM’s development operations, claimed that AIX and SAA will come closer but not converge at a single destination (CI No 1,459). Rather surprisingly, the seminar programme included one Dave Dunsmoir, IBM’s large systems marketing manager, and he talked of the Enterprise Platform. Despite having spent the best part of two decades within IBM’s embrace and the last four developing large systems for the 1990s, Mr Dunsmoir claims to be the ideal person to give an unbiased assessment of the role of the mainframe. His knowledge of the beast goes without question, but impartiality, perhaps, is less certain.

Validity of mainframes

His immediate concern at IBM ’91 was to reassure users who question the validity of mainframes in a world of the PS/2 and RS/6000. He says that RISC processors take existing technology and optimise it to give maximum performance. Unfortunately, some users believe that optimised RISC technolgy will disadvantage complex instruction set processors, and there is growing concern about performance and investment protection. However, this is only true if future technology stays the same, and since technology is moving at a rapid pace, Dunsmoir argues that complex instruction investments are protected in terms of price-performance. For example, Silicon has been written off as yesterday’s logic, with Gallium Arsenide hailed as the way forward. Yet Dunsmoir claims that there is no effective limit to the density of printed Silicon. The constraints don’t lie in the Silicon, but in the photo-technology, and Oxford Instruments’s X-ray synchrotron installed at IBM’s East Fishkill facility is overcoming that. Also, the number of layers that can be achieved in a substrate is increasing rapidly, pushing performance up and costs down. Memory is also becoming less of a bugbear, with 256M-bit chips under development. The mainframe is able to concentrate resource in a single point, and that capability is enhanced by implementing technology such as huge bandwidths, fibre optics and fan or star-like configurations. The 820 and 900 machines are potentially highly parallel, and the same applies to MVS/ESA, and Dunsmoir suggest that if the program space between processors is divided, then the fall off is significantly smaller, which has interesting implications for the handling of enormous amounts of data.