The latest notices for the HP-Intel Itanium chips could be read as another nail in the HP-UX coffin. Long ago, the processor family that powers HP’s Integrity servers ran into trouble, roadblocks that will vex the future for HP's Unix. But it doesn’t necessarily have to be viewed that way. The 3000 gained extra years in spite of people thinking short-term and adopting mass market strategies for enterprise computing. Like the 3000 owners before them, HP-UX owners who’ve migrated from MPE need to think different than a Cheaper, if they can arrange any way to afford it.
What's a Cheaper? That's the manager or consumer for whom the price is the most important concern. They look at today's cash flow instead of the coming five years of ownership cost. They bought $299 netbooks with glee until those slabs of plastic were better suited to prop open windows than run Windows.
You could be a Pricer instead. It’s the kind of pay-what-it's-worth thinking that made the HP 3000 the best value in enterprise computing, at least for value circa the 1990s. So long as HP put its engineering muscle behind a platform that was a walled garden, adding features and embracing new tech, you couldn't buy a business computer that was a better investment than a 3000. When HP bagged its responsibility, the market got left looking for something else. Cheaper looked attractive, after being just stung by a top-shelf expense of dropped futures.
But every platform’s got that day when the futures die in the vendor’s mind. First came Unix, and the promise of everywhere adoption, cheaper than the BMW-caliber MPE. Then Windows, tuned up for running an enterprise with Windows Server and SQL Server. Each cheaper than the last. When Microsoft announced the end of futures for Windows Server and Itanium, MB Foster's Birket Foster pointed out Windows became a lot less cheap since it was made to perform at an enterprise level.
Foster said in 2010, despite Windows Server 2008 being the last version to support Itanium and Integrity, he liked the outlook for HP-UX and the only server which runs it. It all depends, he says, on how far out an IT manager is looking to expect any environment to deliver value. He had a clear view of the lifespan for an OS even then.
"One of my first questions would be, what's your timeframe?" Foster asked in 2010. "How long do you want this platform to be in existence for you?"
After all, customers are not planning out timeframes longer than 3-5 years for any other operating system, so why expect the HP-UX and Itanium picture to run farther toward the horizon? The current HP timeline for HP-UX is through 2025, but it’ll be served up with only revisions of 11i3 HP-UX.
"There are things people can do while they're making their conversions from the 3000 to make it easier to shift the next time," Foster said back then, processes that will make isolation happen. "HP already figured out how to build a hardware abstraction layer so they could run five operating systems on this Itanium chipset. Who's to say you can't build an operating system extraction layer and isolate yourself?"
Something quite like that came to pass with Stromasys and Charon.
Foster said his company did that kind of isolation when they migrated a large oil company off the 3000. And that abstraction layer? OS experts in the 3000 community hoped that eventually, instead of Itanium hosting x86/Xeon programs in hardware, the reverse would happen.
Cheapers may not embrace this choice, since it includes an OS priced for cost of ownership instead of the entry price. They really don't want to consider the extra 25 percent it takes to adopt a better-built, longer-lived product. Not when they can save that money from this year's budget. Pricers think about having to defend their choice in more than five years, instead of looking for another investment to replace one that was never built to last.
Built to Last was supposed to describe HP-UX and Itanium as much as Xeon and Windows Server. But a Pricer needs to know that the vendor will be there for them many years to come, to justify the extra expense up front. Think BMW to consider how much vendor zeal you will need. Can you feel that zeal from your migration platform vendor? Have they spent more in R&D, percentage-wise, than HP does as a company?
Windows will do the job for many migration-bound companies. But the long-term value of anybody else's environment except Linux never seems to be ensured. Even Windows desktop applications get replaced every 18 months. The future of HP-UX is probably not a "we're killing it off" demise like HP planned for MPE.
Instead, Foster says, "In the long run, HP-UX will probably morph into something like Linux." That will be the point when being a Pricer instead of Cheaper might pay off -- because your shop is now full of experts in enterprise-grade IT management solutions, built off the 20 years of the Unix investments, pricier than Windows.