File transfer backup tips flow off HP's forum
HP, Oracle OS promos take hard, soft paths

On HP's Operating Systems, Future and Past

A week ago we spouted off in a podcast about the future of HP-UX, and haven't heard a word over the seven days since about any Itanium and HP-UX recant from Oracle. HP is adhering to its Itanium server roadmap, mostly because the company still sells three OS environments of its own design. Each runs only on Itanium/Integrity servers.

To be accurate, only HP-UX was built by Hewlett-Packard. NonStop and OpenVMS arrived at HP via acquisitions, so HP-UX remains the only OS on today's price list with classic HP genetics. It's also the only Hewlett-Packard OS that HP 3000 customers have used as a migration target. A look back at the fate of HP's operating systems shows a quite of charnel house of OS bones. Only IBM has put down a comparable number of good dogs in bad business circumstances, and Big Blue still supports its business OS's created in the 1980s and even earlier, the AS/400's OS/400 and the System Z for its mainframes.

Alan Tibbetts, a user group director at both Interex and OpenMPE and a 38-year veteran of HP 1000s, reminded us about several of these environments that HP gave up for dead besides MPE. "I must admit that I was amused by the articles saying that HP is pushing a 'private OS' with the webOS product that they acquired from Palm," he said. "Considering that they have killed Rocky Mountain Basic, RTE, MPE, HP-RT, and (any day now) HP-UX, I will have to see how much allegiance they really give to the concept of a proprietary OS."

The difference between things like Rocky Mountain Basic or RTE and MPE is that your OS is poised to move into its second decade running enterprises after HP has quit. But the HP OS history will remind even long-standing HP-UX users that their vendor-supported days are numbered.

Tibbetts even called up a story about HP's proposed replacement for the RTE which drove the HP 1000 servers. HP 1000s were often embedded in other systems, so they didn't have much chance to develop a profile that many independent vendors could embrace. HP followed up RTE with HP-RT, hoping to sell it differently in a marketplace very different from the 1960s when RTE was born. To begin, HP-RT was like today's HP environments outside HP-UX: it was purchased instead of built from scratch.

HP-RT was intended to be a successor to RTE. It used the HP-PA hardware and a real-time operating system that was purchased from Lynx OS and modified at HP. It was marketed differently than RTE, looking for uses as an OEM platform where the sales would be in the thousands or at least hundreds of units. They tried for the set-top box market before that technology became mature enough to be a market. They did convince the Navy to use it, but I don't know how that wound down as they finally pulled the plug on the product.

The problem across the industry is that as microprocessors became ubiquitous, the trend was to push the real-time response much closer to the hardware that was being controlled, down into the first level chips, rather than trying to have a multi-processing system respond to microsecond events. If you have responsive enough chips and sufficient buffering (FIFOs and such) available, then even a system with abysmal response to interrupts (such as Windows) can provide acceptable performance for most tasks. It's only if you have the simultaneous requirements of microsecond interrupt latencies, repeatable deterministic scheduling, and multiple actions which must be tightly synchronized, that you are forced to use hard real-time systems.

So HP was trying to lift this OS into place where independent vendors tied into hardware from OEMs (Original Equipment Makers) to complete a solution. The HP 3000 became HP's first OEM success story, followed by HP-UX about 10 years after the introduction of MPE.

Now HP is saying that it's got a hunger for its own operating environments once more. The message might sound like good news for a 3000 site that plotted its future along the HP-UX course. But the numbers for critical mass might not deliver another 6-10 years of robust development of HP's Unix.

The numbers for webOS, the environment that HP's loving out loud, are not much better in market share. The webOS passion flows from Todd Bradley, HP's leader of its Personal Systems group -- and it's echoed by the new HP CEO Leo Apothker. This is not the sort of forced validation that Carly Fiorina stamped on MPE/iX right before HP announced its exit plans. Apotheker sees webOS as a great business opportunity for HP and its customers.

"We happen to have the greatest operating system currently available in webOS. It's an absolutely outstanding operating system," Apotheker said this spring at OnDemand 2011. "Our TouchPad [tablet] will be coming out in June, and then everybody will see first-hand how good it is. We want to use the webOS to bring everything together."

"We will be putting our webOS on every PC we ship in the next year, as well as on our printers. We see it as a legitimate alternative because it runs devices across the spectrum -- PCs, tablets, smartphones, printers, everything -- and runs them well."

This idea of bringing all of HP computing together started in the 1980s when the company designed the engine for today's 3000s, PA-RISC. HP wanted to drop the redundant development of IO drivers; at the time, HP 9000s ran on Motorola chips, the HP 1000s on another CPU, and the 3000s on a CISC design. Like webOS, PA-RISC was supposed to be a unifying technology. It was built inside HP, however, during an era when HP was fabricating its own chips.

Unlike chips such as PA-RISC, software does not need a manufacturing budget after design; managing yields from wafer platters aren't part of the lifecycle of webOS or HP-UX. HP needs partners for both of these environments, a troubling truth in the light of Oracle's forthcoming pullout from HP-UX. The fastest growing OS's today come from an open source model (Linux) and one built and controlled by a single vendor, Apple's iOS. Apothker believes webOS might give HP a status Apple has long enjoyed: cool.

"If you go back in history, Apple was the cool company with the first PC, the Apple II," Apotheker says. "Client-server was the new wave of computing in the 1980s, and it changed all that we knew about computing at that time. Things are changing again in IT."

HP has long promoted change as a good thing for your industry. It even pumped up an "Adaptive Infrastructure" as way to employ multiple OS's in a single system like the Superdomes. IBM has taken up this kind of design to protect the OS400 environment, letting one box serve that OS as well as Linux, Windows and IBM's Unix, AIX. HP-UX users won't be able to get much more guarantee of the future of HP-UX than they ever got about MPE/iX. However, promoting an OS to a brand and product level may be a promising seedbed for growth in HP's Unix development.