What Size Is Your 3000 Community?
Itanium's static loss is Linux's growth gain

MPE memories grow from Busch's seed

JohnBusch bio It started out with a simple mistake. The popular IT news Web site The Register said engineer John Busch was pretty much the brains behind the HP 3000. Timothy Prickett Morgan, who's had some ill-advised sport with the 3000 before now, thought he'd located the head technologist for your server's magic. Busch, who worked at HP until 1987, "for a dozen years was in charge of the technology behind its HP 3000 proprietary minicomputer platform."

Some 3000 veterans mulled this over for about 15 minutes before they said, "Who?"

Busch sparked some important R&D in the life of the 3000, especially until HP released MPE XL. But IMAGE co-creator Fred White wrote us to say he never stumbled onto a Busch during White's era.

I don't recall anyone named Busch during my years (1969-1980) in Cupertino. Or even after I went to HP Corporate and subsequently to Adager. In any event, I can't see him (whoever he is) as "The Technologist behind the HP 3000." When did J. Busch join HP? Where and in what capacity?

Busch sprouted into view last spring, when The Register was working up his profile to add weight to a story about his startup's new Web caching and MySQL appliances. By this time last week, Prickett Morgan had elevated Busch to "the technologist behind the HP 3000 minicomputer line." We poked into HP's best technical archive, the HP Journal, to verify just how much this engineer contributed. (His official bio appears above.) Along the way in our journey we revisited the glory days of HP's proprietary, elegant innovations. MPE was chief among its software mastery, after a rocky start.

This story is part history and part caution, like so many tales of the 3000 today. The historic part chronicles the massive effort to carry a 15-year old OS into a faster architecture while protecting customers. That's the kind of effort HP performed with gusto for HP-UX, bringing it to Itanium and 64 bits. In 1987, the last time Busch made any mark at HP, the 3000 was getting the same tender loving care as its Unix cousin.

The December 1987 HP Journal issue contained two elaborate articles about the 3000's new MPE XL operating system. Busch was a co-author, along with Al Kondoff, and Darryl Ouye, of MPE XL: The Operating System for HP's Next Generation of Commercial Computer Systems in the Journal. (Yup, you can download the whole issue from HP as a PDF. A very large PDF.) His byline above the largest HP 3000 article ever run in the Journal might be why The Register thinks Busch was "the technologist behind the HP 3000." HP never had anybody called a Technologist on their business cards. It would've been redundant for anyone from the 3000 labs.

The articles from HP conclude out that no one person was behind the 3000's unique technology wizardry, MPE XL.

MPE XL is the result of the effort of over 150 engineers and managers over more than a five-year period. The unswerving dedication of these individuals resulted in the creation of this new production operating system base, which will serve as the foundation for the next decades of HP's commercial computer systems. Each individual who contributed over these years to this operating system should feel a sense of accomplishment and deserves recognition.

I love this secondary headline for the article, redolent of 1980s tech innovation: "MPE XL is a new commercial operating system developed for HP Precision Architecture computer systems. It provides fundamental advances in operating system technology and helps users migrate to the new systems by providing maximum compatibility with existing systems." Even in 1987, HP was working to migrate HP 3000 sites -- to the newer version of the 3000. That same sort of migration is underway today for HP-UX customers, a transition that HP hopes is just as transparent as its designs for MPE XL in 1987.

In that year, the technical theory of MPE XL was being sorely tested in the field. The OS included a novelty called mapped files. One of the engineers who wrote parts of MPE XL, as well as tested it on contract to HP, told us "John Busch and Al Kondoff did create Disc Caching. John was part of the kernel team, but I'm not sure I'd characterize his role as being 'behind the HP 3000 minicomputer line.' "

Some 3000 experts draw a line from disc caching to the MPE XL memory manager, just called "MM" back in the days when knowing such tech nuance could help improve customers' performance. Mapped files made the 3000's processing tap the potential of PA-RISC, another proprietary innovation from Hewlett-Packard. By the time the deadline of the December '87 Journal, MPE XL (1.0) was shipping at last, though to few customers. Many others were calling the deployment of 1.0 "a career decision" for any DP manager -- and they didn't mean a good decision.

But MPE XL survived its x.0 pains and ultimately thrived enough to take over the 3000 installed base, as well as win some new business for the vendor. This technology would have had a better second act if HP had protected the resources of the MPE lab, which grew to unheard-of size while the testing and delayed delivery took place. One rumor had the head count for 3000 development near 1,000 bodies at one point, counting contractors as well as HP's engineers. The rumor might have been preposterous, considering HP's total head count was under 50,000 at the time. But that was a time when business computers were the largest portion of HP's business, so it's possible.

Whatever the headcount, the HP that lifted MPE from a 1970s creation to a 1990s product -- protecting tens of thousands of investments -- now has retreated to the Business Critical Systems group, a suburb of the modest metro area called Enterprise Storage and Servers. BCS takes care of HP-UX, from extending its powers to winning it new customers. Most special of those newest UX users are the HP 3000 sites finding a refuge from HP's departure from MPE.

The caution lies in this part of the story. The glory of delivering that shiny MPE XL sounds about the same as HP's story about current HP-UX supremacy over Unix designs from Sun or IBM. Busch left HP and returned to his roots as a Stanford computer science professor for awhile, but eventually helped research Sun's RISC designs before he started Schooner Information Technology. The Register cares about Schooner because, like Prickett Morgan, it's got deep roots into IBM by now.

But allowing any IBM acolyte to tell the 3000's history in such inaccurate shorthand seems another underserved slight for such a storied system. The warning in this lesson about proprietary wonders like the MPE XL emulator: They're still proprietary creations. Anything proprietary needs a stalwart steward of a vendor, because the flame of faith about such a product's future burns hottest in the vendor's labs -- and management offices. As it turned out, "the next decades of the operating system's base" in that headline turned out to be less than two. Unless you win and protect enough sales to create a critical mass, there's no market to help pass a shiny proprietary torch to fresh generations.