NonStop - A Running Commentary - November, ‘08

The opinions expressed here are solely those of the author
and shouldn’t be viewed as reflecting any views
of his current or past employer

I have been catching up on my reading this week and I was a bit surprised by how far I had fallen behind. Along with skimming the headlines, I always check the major advertisements as a kind of “litmus test” on those subjects being viewed as topical by the major vendors.  Imagine my surprise when I came across an IBM advertisement in Fortune Magazine (Nov 24^th, ’08) that heralded “Decrease Costs. Increase Growth.” Wow!

Could IBM be onto something here? It’s hard to believe that this has only just come to their attention, and even more surprising to see IBM highlight it in an advertisement. IBM then goes on to add “Legacy apps. Faulty data. Redundant silos. In banks all around the world, millions of dollars are tied up in outdated, patch-worked systems.” And then it hit me, this has to be a message for their customers - who else could they be addressing but their own user community? After all, it’s been more than four decades of overlapping systems, with highly confusing product roadmaps, that have contributed to the “patch-worked systems” that IBM sees deployed. And are they only now surprised?
 

There was a time when IBM faced the prospect of being split into three independent companies. According to papers I found in the Hagley Library, a group that collects, preserves, and interprets the unfolding history of the American enterprise, the success of the System/360, and the practice of ‘bundling,’ led to the US Government filing charges on January 17, ’69  “…that IBM had hindered the development of service and peripherals competitors by maintaining a single price policy for its machines, software and support services (bundling) … and that IBM had introduced underpriced models knowing that they could not be produced on time and did this to prevent the placement of competitors' machines.”

For those readers interested in the full story, check out the paper: http://www.hagley.org/library/collections/manuscripts/findingaids/ibmantitrustpart2.ACC1980.htm

Uncertain of the outcome of these filings by the US Justice department, IBM quickly created the Data Processing Division (DPD), General Systems Division (GSD) and Office Products Division (OPD), with each of them offering their own computer lines. There was no way that overlapping technology silos wouldn’t develop, given this organizational structure and the competitive nature of each division’s sales force.  

 

And compete they did.  Ellen Hancock of the IBM Communications Group, and part of DPD, introduced the Series/1 minicomputers that were aggressively sold to banks as ideal branch concentrators. And then DPD developed the 8100 minicomputer to compete with the S/36 and S/38 minicomputer from GSD. As the ‘70s finally came to a close, DPD even added a small, distributed mainframe – the 4331 / 4341 – into the mix, for good measure!  

Entering the ‘80s, data center managers that had stayed loyal to IBM found themselves overseeing mainframes running MVS, surrounded by System/1 communications computers, networked to 8100 minicomputers, as well as System 38’s, with their tightly integrated data base centric stack, and even regional, distributed 4331 mainframes running VSE. Not forgetting the 3600 / 4700 in-branch controllers that were essentially separate lines of minicomputers commonly deployed by banks and retailers. Altogether an ugly mess that no measure of lipstick could ever disguise!

In other words, IBM business plan of the ‘70s and the ’80s was instrumental in the emergence of redundant silos – in part because it wanted to make sure the company had multiple viable product lines should the business be divided into three independent corporations. But driven, just as much, by the competition between the divisions that was left unchecked by a succession of IBM CEO’s which were reluctant to reign-in and rationalize the many product offerings. 

When the Reagan administration came to power, the suite against IBM was dropped and the justice department elected to keep IBM whole, and break apart AT&T instead – an outcome that caught many industry watchers by surprise. From the Hagley Library papers referenced above, “William F. Baxter, as the assistant attorney general in charge of the Antitrust Division, Department of Justice, ‘82 signed a Stipulation of Dismissal on January 8 that stated the government's charges were ‘without merit.’ It was later discovered that Baxter had failed to disclose that he had been retained as a consultant to a West Coast law firm defending IBM in private antitrust cases.” Ooops!

Dodging a major bullet and grateful that it could continue in one piece, does IBM then restructure and rationalize its product offerings? Well, actually no! It seems to have fallen in love with having many overlapping product lines – with each of these products supporting an overlapping mix of operating systems, transaction and application processing infrastructures, and data base management systems. Faulty data? Redundant silos? Should we be surprised given IBM’s history! And should IBM users view this as anything extraordinary?

One of the more impressive aspects of the NonStop architecture has been it’s scalability and data integrity – two very key attributes of NonStop that are so often overlooked. The energy expended by the development teams within NonStop in the design of the hardware and microcode, in support of redundant paths and data checking, to make sure data is never corrupted cannot be overlooked these days. So, faulty data is not a charge you can level at the NonStop user.

And even more important to me is the issue of scalability. No matter whether the need is for an expand-connected four, or six 16 CPU cluster or a single 4 CPU node, the architecture, infrastructure and manageability remains the same. Add another couple of CPUs, or even a complete node, and it can be done without interruption or an outage of any kind. So, the ability to size by transaction demand gave many companies the option to size NonStop servers differently per location and lessens the need to add different technologies or solutions to the mix. No redundant silos to be found anywhere with such an attribute.

Overall, I give HP good marks for reigning in the potential explosion in product offerings following the acquisition of Compaq. For the NonStop community there was certainly relief to see the Integrity line include support for NonStop. With the arrival of blades, a quick check of the HP web site lists support for NonStop and OpenVMS alongside of Linux, Windows, and Unix. And I am convinced that further blade rationalization will appear in the future which will drive hardware pricing even lower.

IBM concludes it’s advertisement in Fortune by stating that “with a wide range of services and technology offerings ... IBM can manage your entire (application) operations or help optimize individual parts of the process.” But comes up short of identifying exactly what mix of hardware or software it is promoting – and again, leaving the door wide open for legacy apps, faulty data, and redundant silos remaining with us indefinitely!

Today, IBM is a technology services and software company, and perhaps there never will be any attempt to reduce the complexity inherent with its product offerings. After all, you have to wonder if IBM would ever back away from offering services directed at its user community left with the legacy apps, faulty data, and redundant silos that they have invested so much energy in producing for close on half a century! It’s a risky and somewhat flawed strategy but they seem content and oblivious to the absurdity of it all. Quick, I think we need more lipstick … 

Richard Buckle

HP – ITUG:            Board of Directors (2000 – 2006)    Chairman (2004 & 2005)

IBM – SHARE:      Board of Directors (2007 – 2008)

Real Time View blog:           www.itug-connection.blogspot.com