On HP's Operating Systems, Future and Past
LinkedIn 3000 group grows, opens its posts

HP, Oracle OS promos take hard, soft paths

Over the weekend HP ended an Itanium promotion program designed to win customers for HP-UX. For six months the vendor was giving away $12,000 Integrity servers to any company willing to try out HP's Unix. Analysts said the deal was HP's play to lure Sun's Unix users away from Solaris. Sun's answer came after HP's former CEO took over Sun hardware and OS operations. Oracle, the owners of Sun, simply won't support HP-UX beyond version 11.

Actually, Oracle said it wouldn't support Itanium servers on Oracle 12. But since HP-UX runs only on Itanium, the Oracle Itanium ultimatum amounts to cutting off HP's Unix from its most installed database. Oracle is betting HP-UX won't migrate to x86 hardware. It's probably a safe bet.

HP took the hard-ware path to pushing its Unix, while Sun-Oracle shot a soft cannonball across the bow of Good Ship HP-UX. One was a carrot, the other a stick. Since some users began to ask if HP's creating a port of HP-UX, Kristie Popp, a social media rep in HP's Enterprise Server, Storage and Networking group has replied, "Please note, HP has no plans to port HP-UX to x86."

The HP 3000 faced a similar crossroads 10 years ago this spring. At that time Itanium chips were the early-decade darlings of HP's enterprise futures. HP-UX was already ported to run on Itanium, whenever HP could get its act together and ship servers driven by the chips. HP "had no plans to port" MPE/iX to Itanium hardware. Users heard at the 2001 Interex conference that PA-RISC had plenty of headroom for years to come. One HP 3000 leader in the community believes HP should be preparing its Unix users for a transition to Linux. Free hardware promos won't keep the HP-UX business alive.

"HP should be trying to buy SUSE Linux," says Donna Hoffmeister of Allegro Consultants, where they support both MPE as well as HP-UX. "If they want to offer a soup-to-nuts Linux-oriented solution, that is the way to go. I'd be utterly delighted to see HP do something to keep and attract folks to HP-UX. One might think that they'd figure out that it takes capital investment to keep an OS alive. They certainly have recent experience with the results of the inverse of that policy."

HP's 2001 assurance that no investment was needed in MPE/iX-on-Itanium should give any migrating system managers pause if choosing HP-UX. The x86 port for HP-UX looks like an even longer shot than porting MPE/iX to Itanium in 2001. Just like HP's message of a decade ago, the new Unix message is that fresh chips don't matter to an enterprise.

It will take years for the ripple of Oracle's decision to swamp the HP-UX boat, especially since some shops have put Oracle on the x86-based server while the applications run on Integrity systems. But customers already feel like T-Rex during that era when it just kept getting colder.

"Are we headed the way of the dinosaur? Will we one day be Solaris?" asked Unix admin Court Campbell, administering HP-UX and Oracle 12 at a Houston employer. Campbell noted that Red Hat and Microsoft have also pulled away from Itanium. But those moves make more sense since they come from operating environment vendors.

When Campbell asked in LinkedIn's HP-UX forum if there's any truth to rumors about moving HP's Unix to x86 to create "HP-NG," Olivier Masse of Hydro-Quebec expressed serious doubts.

"I don't see why they would do this," Masse said in a posting. "I think they'll milk the cow for as long as they can, catering to current enterprise customers, and call it a day when it will become not profitable enough to keep all the R&D and support infrastructure for HP-UX.

Hydro-Quebec has already seen its exit from MPE/iX once HP cut off hardware futures for the OS. They're now running on two Itanium-based environments, HP-UX and OpenVMS.

"The x86 platform is already saturated and it would cost a lot of dough to port HP-UX outright to x86," Masse said. "The Linux kernel is more modern, and has benefited from R&D from both academia and the private sector for years. I don't see new customers flocking to an 'NG-UX' anytime soon."

Customers see the adaptive infrastructure HP promotes as including more than one processor for HP-UX. It looks like a gaping hole to Charles Ruedi, the senior Unix system engineer at Almac Clinical Services. "The reason HP should have done this long ago was so customers who wanted to do a proof of concept or something similar didn't need to purchase an expensive itanium box to test with," he said. "Also, it would be great for the development space."

HP's got history in supporting software such as HP-UX, but the veterans from the HP Way days don't recall the legend as a positive note. Tom Lang, who worked at HP's Office Products Division in Pinewood when the company was breaking ground with HP NewWave. "They purchased Compaq," he told us, ,"then they abandoned MPE. Then they purchased a hand-held [company in Palm]. Now they claim to be in the software industry?"

I was working in the internal EDP department at HP Pinewood when they were developing HP Office about 1987. What a mess that was! Every OPD developer had their own 3000 to develop and test their piece of the action. Never did it cross their minds to test the full product on one machine. I foolishly tried to point this out, suggesting that the verbose product required a lot of disc and memory, notwithstanding that it was a CPU hog. Management didn't listen, and anyone who remembers this clunker uses it as an example of why HP is not a software company -- hard as they may try.

Then they tried NewWave. Then they purchased the Apollo group, acquired a version of Unix and label-engineered it as HP-UX with HP Workstations. Then they offered the HP 9000 as a product. Unix was developed by Bell Labs as a means to speed-up testing on an IBM mainframe. By removing all the security in the IBM OS, they were able to develop faster, then do full testing on the full OS. Everyone wanted a copy of Unix, Bell Labs provided, and the cat was out of the bag -- operating systems with no security. Textbooks were written about Unix, they taught it in colleges and universities, but nobody thought to warn business users that they were taking on a business system with no security. But it sold, and it continues to sell.

Everyone now scrambles to develop versions that have some form of security. But a reasonable person knows that if it's not there at design-time, it'll take a bigger effort to try and put it in later. Unix textbooks today still do not have sections/chapters dedicated to security. It is understandable that 'security' is not fully explained, but no mention of it, not even in the index, is self-explanatory.