Rising Sun, setting Unix: HP's next migration unfolds in secret slides, emails
May 18, 2012
Ever wonder what the demise of the 3000 inside HP looked like? The event that reshaped all of our careers surfaced suddenly for some. For other community members, the vendor's departure was inevitable, given the indicators they followed. The week the US courts lifted the inevitable veil off HP-UX. Hewlett-Packard used its business acumen to decide the lifespan of its 3000 business. Now we can see what that kind of review looked like, thanks to Oracle and a fired HP CEO.
There is little explanation for how Oracle knew which secret emails and slides to uncover but one -- Mark Hurd and his leave-behinds at HP had these maps in hand. They knew exactly what to request in the discovery phase. It's unprecendented, to my eye. I saw an HP purchase order for $22 million per quarter paid to one vendor. If you wonder what something like an $88 million annual PO looks like, click on the graphic above. HP was spending like this for years, all to ensure that Intel would keep developing and creating Itanium processors. It wasn't spending anything to migrate HP-UX to a non-Itanium, commodity chip. Before long, these Unix customers -- plus ones using VMS, NonStop and more -- will do that migration instead. Linux on Intel. I can't even guess what NonStop or VMS will do.
These are the heart of HP's remaining proprietary computing environments. NonStop, OpenVMS and HP-UX use Itanium as crucially as a liver in a human body. Pull out Itanium from HP's futures and you have no more reason for any customer to leave their apps on these operating systems. Because the OSs don't run anywhere else. HP knew this and talked about it, both in its internal meetings as well as high tension negotiations with Intel. It's just that HP was saying something very different to the public. So was Intel. Anybody who believes Intel has other ideas about Itanium futures needs to read a few of the released emails.
If you don't have time for that, just scan the PowerPoint slides. There's a stunning one below from 2007, mapping steep declines to zero for the Itanium computers. (Click it for details.) You can look at the "Blackbird" proposal from an exhibit, too -- the one where HP sized up the pros and cons of buying Sun. (View the Blackbird)
A reporter from All Things D, the tech website run by the conservative Wall Street Journal, posted these emails and slides that were once secret, but now released by the court hearing lawsuits. Arik Hesseldal's article is must-reading for anyone who needs to plan an IT architecture or report on futures to CEOs or VPs of Finance. Hesseldal sums up HP's own view of the future of the company's only single-vendor 3000 migration target.
Key phrase: HP-UX, its version of Unix developed specifically for Itanium servers, “is on a death march” because of Itanium’s inevitable demise.
Why care, if you're already migrated off the 3000? It's as simple as an ostrich. If you've put your company's money on the HP-UX platform -- and think it's got a good run left in it -- you're hiding in the sand. It pains me to have to acknowledge anything that Larry Ellison's Oracle asserts. But there's no other reason to believe this won't work out the same as the 3000's evaporation off HP PowerPoints, strategy statements or price lists. The end is more than near. It's nearly here.
Update: HP's also dropped its own stink-bomb of documents, later in the same day, several emails plus pages of text message transcriptions between Oracle salesmen and execs. Most notable: an email from Lorraine Bartlett last March, just days before Oracle's pullout from Itanium. Bartlett, VP of Marketing for the HP-UX host Business Critical Systems, is effusive in praising her company's message about HP-UX futures. HP's "Kinetic" strategy, shared with analysts in March that was "a big hit, and really resonated," included messages about "HP-UX unbound" and a common socket design that Intel was announcing to give Itanium chips the same underwear as Xeon chips. The texts between Oracle sales people and managers have a college frat-boy tone to them -- but seem to be in HP's bomb only to show that Oracle knew the HP-UX competitor Solaris was "a pig with lipstick." (Warning, salty language there.)
At least now our community's customers can now see examples of the language and philosophy and schemes that made up the 3000's departure. "Don't possibly signal to world end of roadmap..." versus "We'll have roadmap updates in the future." A product relies on growth from the outside market, plus the technology becomes too costly for HP's budget. That's the 3000's story from HP's view. No different, except in number of customers, from today's Itanium story. Five years ago HP worked up an estimate of the price to move HP-UX to the commodity Xeon chips. About $100 million, it learned, to make the Itanium dead-end go away. But HP opted for a $88 million per year alternative with a short future for its commodity environments. It propped up the chip instead of reinvesting in development its own OS products. It made those decisions while its CEO slashed R&D budgets below the bone.
And that CEO continues to determine the future of HP-UX, even after HP fired him. See, Mark Hurd got himself hired by a company working to kill off HP's Unix. Larry Ellison called the board's ouster of Hurd -- after Hurd's creepy and sad debacle of chasing a reality TV actress, instead of his wife -- one of the worst decisions HP ever made. With the release of these secret emails, it looks like HP made a decision even worse. To a customer who uses HP's Unix, VMS or NonStop, HP never should have let a competitor in the Unix market hire Hurd.
A few months ago a respected tech icon in the NonStop market wrote about the future of HP-UX. Dr. Bill Highleyman thought that the forecast which I'd offered on Itanium was dubious -- that announcing an Odyssey project to get the best of HP's Unix onto Linux meant the end of Itanium, therefore also HP-UX -- and it was nearby. I would invite Dr. Highleyman, plus anyone in our community who remembers losing hundreds of thousands or even millions of dollars of MPE business, to have a look at the archive of documents that's been curated by All Things D.
Some of the least fortunate customers will now have to migrate away from HP's Unix. Or they can live in the fine-tuned OS afterlife beyond HP. Given the health of Hewlett-Packard's business these days, maybe that post-HP afterlife will seem more lively.
At the least, life in the afterlife will honor the economic advantage of an OS built for a chip the vendor owns, like MPE and PA-RISC. Unix planners are being invited on an HP Odyssey to commodity computing. How anybody can cost-justify that journey, instead of a genuine commodity solution -- well, that feels like a well-kept secret. What's going to be out in the open in the lawsuit trial is more muck, and murk, around the genuine future of the last proprietary OS that HP's ever going to build.