SAA: IBM's vision of the future

Systems Application Architecture

Ideas & Trends

Henry Fersko-Weiss
Lotus

April 1, 1988

When IBM announced Systems Application Architecture in March 1987, it was not unlike a school of painters announcing a new artistic movement. SAA is nothing less than a manifesto--a statement of much of what is and what isn't crucial to IBM's vision of the superstructure of computing at all levels of hardware. And like an artistic movement, it will have its adherents, detractors, and renegades.

Because SAA is a set of programming conventions, protocols, and standards, it is not something that people sitting in front of a PC will have to know anything about. In fact, most PC users will probably never even hear about SAA, just as most users have never heard of SNA (Systems Network Architecture), a set of IBM conventions in use for the last 13 years that dictates how systems communicate to each other. However, even if users are unaware of SAA, they will experience its impact.

Every program written to SAA standards will have four components in common: a user access, a communications support, a programming interface, and a set of applications. SAA is an attempt by IBM to bring a consistent look and feel to programs. "Over time SAA will serve as a framework for developing and using the same applications program across the major IBM computing environments, from the System/370 to the System/3X and PS/2," says an IBM spokesperson.

What this means in everyday terms is that PCs, minicomputers, and mainframes will be able to talk directly to each other, to communicate without the awkward machinations now necessary on networks. Up to now these three computing environments were isolated territories, with different operational techniques and protocols.

"If you are a PC user and you also use mid-range and mainframe computers, you have to have different training and a different mind set for each," says Jonathan Yarmis, a research analyst at the Gartner Group (Stamford, Conn.) "It's grossly inefficient to have to change the way you use the keyboard or maneuver around in a program as you go from one hardware platform to another. SAA is an attempt to bring consistency to all IBM equipment so that users don't have to be aware of the hardware they are using."

For years IBM customers have been asking for direct communications between levels of equipment. It may seem like an obvious goal, but IBM is at heart a mainframe company and has had difficulty in granting equality to minicomputers and PCs. "SAA is a recognition of the problems IBM has had with multiple tiers that are incompatible from both software-development and user-interaction standpoints," says Yarmis.

It is also a recognition of the success of companies like Digital Equipment Corp. and Apple Computer Inc. DEC's applications and network architectures are absolutely homogeneous, from the smallest MicroVAX to the biggest VAX cluster. This is one of the main reasons DEC has been making inroads on IBM for years. Apple has been getting its Macintosh computers into corporations by using the same philosophy in developing or fostering applications software. Programs for the Mac have an appealingly consistent user interface.

Now IBM is shooting for the same level of consistency. To do this, it is anointing only certain programming languages. For applications development, only C, Cobol, and Fortran have been chosen--so far. In the database area, Structured Query Language has become the one holy tongue (see story on SQL database engines on page 11).

Other languages may be included as time goes on because SAA is not a static standard. It is meant to be an evolutionary structure, like SNA. In addition to singling out specific languages, IBM has described how screen menus, icons, and dialog boxes should appear. It is at the user interface that you will see the results of SAA. And it will have its biggest impact on large networks where PCs interact with other computers.

"SAA is really for networks," says Clare Fleig, director of research for the International Technology Group, a computer-management consulting company in Los Altos, Calif. "It will have its impact on single systems, but the real benefits will be on networks, where there will be a common way to work and talk with different systems."

Before SAA, IBM tried to reshape SNA into an architecture that would solve the consistency problem. But SNA principles predate personal computers and major evolutionary changes to minicomputers andmainframes, so the solutions became more and more cumbersome as time went on.

"From 1979 SNA has become increasingly complex," says Fleig. "SAA includes strict programming rules that will eventually translate into how menus and help screens will look to the user."

For users SAA will provide systems that are easier to learn and use. Because the environment will be the same, or nearly the same, for different applications running on different types of computers, the learning curve won't be very steep. Plus, many of the procedures in retrieving information from different systems will be done by the network, not by the user.

It will certainly take time, however, for SAA thinking to sink into new software offerings and for IBM to remold old software. Unlike an artistic manifesto, SAA is not a brief statement of a dozen or so principles. The first set of specifications is about 200 pages long, with a 100-page entry guide, and that's just the beginning. There will be lengthy extensions in the future.

According to Bill Kirwin, also with the Gartner Group, "It will take a decade for an entire SAA portfolio of programs to roll out."

While IBM is rolling SAA out, the world will not stand still. Kirwin thinks that other computer companies may develop higher-performance integrated environments and more powerful computers for less money. "Some shops will opt for earlier solutions from other vendors. There is evidence for this already in the recent popularity of the Macintosh in the office."

Lost time in waiting for SAA is not the only problem MIS directors and users face. SAA is still a somewhat awkward fix to hardware systems that were developed without direct connectivity in mind. As a result, users will need faster and more powerful equipment to cover up losses in operating efficiency.

Despite these costs there is little doubt that SAA will have a huge impact. Over time, software applications will become more consistent, and the process of passing data between mainframes, minis, and PCs will become much easier.