HP/DEC/Compaq

Interview with HP CEO Meg Whitman.

http://allthingsd.com/20120605/hewlett- ... ot-to-say/



http://www.oracle.com/us/corporate/feat ... 623014.pdf <-- also interesting...

yes, we absolutely need another derivate on x86, because there is so much to sack up in this area

Almost all of which are the same old thing with a different sticker on the front. If HP spends the time and money to really do something first-class then there could be a definite reason to have it. NonStop AMD64 hardware could definitely have a market, something built really well running OpenVMS, sure.

I do admit that, of all the HP proprietary O/Ses, HP-UX is probably the least compelling of them to move over. x86 already has xBSD and Solaris which are both good UNIX systems (excluding Solaris' licensing from the discussion). Sure, HP-UX is fairly solid, but it's really nothing special or new in the x86 world.

Okay, I kind of work for HP, and there was a massive facepalm when reading this very misleading article with wrong conclusions.

Just to set things straight...

There will be no HPUX on x86. Never. Ever.

The port was started in the labs many years back, and it got up to a state where you could actually boot it. But it was killed, and for a few good reasons:
- Customers don't trust x86. Really, they don't. x86 is nice as an entry-level commodity platform, but that's it. Yes, Xeon has been incorporating many of the resilience features of Itanium by now, but there still are a lot of them that simply don't exist on x86. And it's not only the CPU, it's also the machine itself - chipsets, IO, you name it.
- x86 can't handle the load anyway. It only scales up to four sockets, and even then. Cache snooping is bad.
- It would cost many millions of dollars to make the port production-ready. $100 million, according to the estimates. And we all know how things go with estimates
- There will be no ISV support anyway. Why bother spending the above $100 million on an OS that boots, but... there still is no Oracle DB on it? I mean, yes, Oracle killed Itanium support, thereby targeting HPUX. But do you really think Oracle will jump for joy and develop a new port of their DB to a new competing platform? Larry isn't crazy.

So, in short; no, there won't be HPUX on x86. It would be corporate suicide.

The Project Odyssey that Whitman talked about is a project which aims to bring the Integrity line features to x86, as much as possible. Ultimately, we want to have x86 hardware that is as good as the Integrity line in any way, except for the actual CPU chip. And these servers will run Linux (or Windows. But heh. Yeah. Windows.). Unfortunately, for management, the difference between Linux and UNIX is nil.

I'll bet my left nut that what Whitman wanted to say was " Ultimately, we've got to build a Unix-class system on a Xeon chip" . Not HPUX. A Linux, but Unix-class for reliability and performance. HPUX will continue to run on Itanium, and contrary to Oracle claims, that chip is pretty damn good and not nearly EOL yet. And if at some point in the future customers do pull the plug on their UXes because of upgrade paths for eg. Oracle, the Odyssey machines should be there to take over. If the interviewers at allthingshd would have bothered to even check the stuff they're writing on, they would have known this, and I wouldn't have to spend my time explaining customers (and nekochan ) that some misinformed dimwit misinterpreted a CEO's somewhat-too-fuzzy talk on a delicate subject.

Sorry to spoil the fun - I would have loved to play with UX on x86, at least as much as anyone here. But it's not going to happen.

I would love to see OpenVMS on a well-built Integrity-class system running on the AMD64 architecture. Frankly, if HP-UX died tomorrow, it wouldn't really be that much of a loss - sure, there are some nice features, but nothing that makes it truly unique given the competition. VMS is a wonderful system that really is totally different to anything else out there. It'll never happen, but a VMS renaissance would make me very happy

If I'm reading the Oracun disclosure case info's correctly, it appears that there "was" a project that got HP-UX to IPL on x86 to a rudimentary userland level and it was cancelled prior to staff being for the high-jump and everything moving to India.

I'd kill to see VMS on x86 (Purely to save it.. because I don't see them ever bringing back the VAX). The port to alpha shows pretty clearly you don't need exotic priv levels in your hardware any more.

How do you figure? The Alpha is not what I'd call "exotic", but it is significantly richer from an OS support standpoint than its erstwhile competitors.


[Digital Technical Journal Vol. 4 no. 4, §7]

Sorry.. I ment IA64.. "alpha" indeed.. brainfart. IA64 doesn't have the "hardware assist" modes of VAX/Alpha so we at least know VMS on x86 is possible.

Yup. But being able to IPL is still a long way (and a fuckton of money) off from being production ready. No surprise it was cancelled...


Yeah, don't count on that either. You can run VMS in emulators easily, though, so at least it's not entirely doomed for hobbyists. I have three SIMH emulated VAXen running in a cluster, just for fun. And the new hobbyist program works *really* well, they even distribute pretty damn recent media kits electronically (for free!)

True..there will always be emulators. I though am one of those odd individuals that really enjoys restoring, managing and dealing with the actual hardware and issues, reading about things.

Emulators have always felt like "cheating" .. but I'm sure there'll be a day when getting the hardware will simply be impossible.

You can still get Itanium easily. Powerful enough to do proper work on, too