You can't blame Hurd for being inconsistent. almost.
He fucked* HP Employees, Shareholders and Customers alike.
(* Though apparently couldn't fuck a particular contractor which led to all this mess).
Hindsight is always at least 20/20. But sometimes it is 20/15 or even 20/10. Maybe Hewlett-Packard should have bought the server and storage business from Sun Microsystems and argued to keep Solaris and let Oracle eat Java, which is what the company really wanted to get ahold of once it became clear that IBM was not going to buy …
He fucked* HP Employees, Shareholders and Customers alike.
(* Though apparently couldn't fuck a particular contractor which led to all this mess).
...at the hands of Hurd, and had a difficult relationship with Oracle. I find it quite amusing to see these two, rather unsavoury companies at each others throats. Lots of blood and guts please :-)
...but that's because I'm an engineer. It looks pretty plain that they won't though - Oracle have plainly planned this out with their lawyers, and wouldn't have kicked off unless they were 95% sure of the full-time result.
Score 1 to the shysters. Oracle doesn't do retail, so doesn't have to care much about Joe-public image. If everything works out right for them, they may even be able to eat HP.
As per the title.
It's clear Itanium is a dying breed, and Oracle simply joined the likes of Red Hat and Microsoft in dropping support for the architecture, so all HP possibly have to complain about are any possible promises or contracts made by Oracle, to support the platform in the future.
In case you didn't know, the engineering bits that once were part of HP are now called Agilent, and they have nothing to do with this spat. The purveyor of printer ink and assorted bits of computing hardware claiming the HP name has little affinity with engineering, and the only thing it invents (to try to put some credibility into their slogan) is ways to sell pigments for prices that exceed gold, weight by weight.
I hope both HP and Oracle lose
While HP might be a really sucky imitation of a once great company, it doesn't mean that it's only "peddling ink" and PCs - that's a very retail consumer centric view. In fact this article has nothing to do with printer/pc/tablet division of HP.
Their enterprise division peddles big iron in the same market as Netapp, EMC, Sun, CISCO in the enterprise world (not too successfully, one would guess), but they do have an engineering which R&Ds these products and this particular article is in relation to the arm which competes against Sun & IBM for UNIX server business... which a common Joe might not care for - but it's still a 20billion $ business.
I know HP deals in more than ink; I deal with a segment of their "enterprise division" hardware and its associated OS daily.
Given their effort in supporting that platform, I'm fully justified in calling them a purveyor of pigments.
Just curious: what HW/OS would that be ? (Assuming they wouldn't be able to single you out based on the data! :).
"It's clear Itanium is a dying breed....." So please explain why Oracle sell more licences for their software for Itanium than for their own SPARC designs? Or, why IBM Software sell more software licences on hp-ux than AIX? By your reckoning, if Itanium is "dead", that means SPARC and Power must be fossilised!
".....Red Hat and Microsoft in dropping support...." Completely different case. Both RH and M$ couldn't make enough sales on Itanium because us customers preferred hp-ux. They also hadn't entered into the same sort of agreements as Oracle had. Try a little research next time.
The one that the totally-not-engineering-minded Compaq (*spit*) killed a perfectly fine processor for, so as to score another platform for the "industry-standard" Itanic.
So I will go with AXP and the truly 64 bit one, eh ? Can't blame you for feeling left high and dry.
"Maybe Hewlett-Packard should have bought the server and storage business from Sun Microsystems"
Am I the only one to think that for either IBM or HP to have acquired Sun's hardware would've raised a very real (and huge) red flag for the competition authorities? They took enough interest in the red herring of Oracle and MySQL!
It's ugly and the only winners are IBM and future IBM customers.
Too many customers are lazy and don't adopt better technology. This is certainly a perfect opportunity to justify the move to IBM's Power systems.
It's funny how HP complains about how Oracle double the price on Itanium with no business or technical justification. What they leave out is Oracle finally moved it up to where Power6 and Power7 has been for years. IBM never complained...maybe that is because Power systems deliver 2.5 times more performance per core than Itanium systems?
Since when is a press release an implied contract?
Funny to read how HP complains about Oracle having a better relationship with IBM. Maybe HP should talk to SAP about buying Sybase ASE. SAP just wanted the mobile software and got the DB.
Looks like customers don't have to wait till July 2013 to feel the pain of Oracle's abandonment of Itanium.
Cheers Matt B......a.k.a. Henrik Elmgren
"....the only winners are IBM..." Maybe, but I think SAP and some of the ProgreSQL people will be quite happy. The hp ProLiant team are probably going to make hay too. IBM? They might get lucky with those customers that don't realise Larry could do exactly the same to Power, and IBM wouldn't have the legal recourse hp does to fall back on.
"......Too many customers are lazy and don't adopt better technology....." I know, so many of them stay on that expensive Pee-series stuff, isn't it awful? I mean, have you seen their Pee7 blades? They have to cripple the cores in order to keep within the poor power and cooling capabilities of the IBM chassis. Tragic.
"....maybe that is because Power systems deliver 2.5 times more performance per core than Itanium systems?....." Only in the IBM labs. Everywhere else it seems that hp win more of those enterprise orders, going on the average value of the Integrity vs Pee-series sales shown in the last quarter's results. You don't seriously want to pretend that hp could sell more high-end systems than IBM if the IBM kit really could go 2.5 times faster, do you?
"....Since when is a press release an implied contract?...." If you're referring to the hp-Oracle Alliance Agreement, it was a contract with plenty of money involved. Oh, but then you would actually have to know what you were talking about to know that!
".....Maybe HP should talk to SAP about buying Sybase ASE...." Oh, what a good little IBM troll, you managed to work in a dig against Sybase too! I'm sure you'll get a few extra peanuts as a reward if IBM sell some more DB2 licences. But you forgot to mention that DB2 is supported with hp-ux on Integrity, even with SAP. It's our second choice to Oracle with hp-ux on Integrity with SAP. AIX and Pee-series is third. Don't be too upset - you at least beat the Slowaris-SPARC offering.
"....Henrik Elmgren." ?? If you mean the Henrik Elmgren that works for hp then he is in California, whereas anyone with even the attention span of a mayfly would have noticed I've already admitted to being UK-based. Try again, little troll.
Why does HP carry on with the Itanium processor? They should have swapped it for XEONs, or even Opterons on the Integrity platform, then no issue, and better price/performane platform.
I think Oracle has a point in not wanting to support a platform with zero future.
As pointed out by others repeatedly, HP-UX, NSK and OpenVMS all run on Itanium (in the case of NSK and VMS, ONLY on Itanium for new hardware). There are very good reasons to want to run these OSs and those are way more important than the religion of x86 etc.
The PP is largely irrelevant for the systems that run Integrity.
If you hadn't noticed, hp is the leading x64 vendor with the ProLiant range, and as part of that also the leading Linux server vendor. Integrity gives hp the ability to attack IBM on two fronts - the Wintel/Lintel front and the UNIX front. You could even mention a third front if you want to talk about NonStop taking sales away from IBM mainframes. Snoreacle is really too minor a server vendor for hp to worry about. All three fronts are very profitable and not at all dead, despite what Larry wishes. As to having no future, the Itanium has a much longer and more detailed roadmap than Power, and a much more believable one than the Snoreacle CMT one (don't even think about the SPARC64 roadmap as it is essentially dead after the next speedbump). Sunshiners spent years telling us Itanium was dead over and over again, and then Sun died. Larry is now the proud owner of the Sun server zombie, so I suppose it's only fitting he should be maintaining the Sunshiner tradition of talking out of his rectum.
HP-UX used to run on PA-RISC, and was ported to Itanium.
NSK used to run on MIPS, and was ported to Itanium. And before that NSK ran on multiple proprietary CPUs, and was ported to MIPS.
VMS used to run on Alpha, and was ported to Itanium. Before that, VMS ran on VAX, and was ported to Alpha.
There is no reason these operating systems cannot be ported to x86, especially the Westemere EX processor.
Also, there is considerable hardware design sharing in HP. The HP C7000 Blade System was based on technology developed by Compaq/Tandem. The Superdome 2 is a hybrid of the C7000 and the original HP/Convex NUMA interconnect. With the Itanium 9300 and Intel 7500/E7 sharing Intel QPI technology, the hardware would not be a barrier. Notice how SGI was able to design the Altix UV to use Nehalem-EX processors.
I don't see much reason to port HP-UX to x86, but certainly NSK and VMS could be ported. Then HP could work with Red Hat to qualify RHEL 6.x on a Westmere based Superdome 2.
Presumably they are giving them away because they are Sun servers that no other bugger wants either.
Carry on lads, I'm enjoying this..........
I keep on wondering why Intel is not behind HP and it let's the former complain about Itanium. If Oracle was actually wrong and Intel is not going to EOL the Itanium platform, then it would have gone full straight to combat Oracle. Now, all we have seen is just a press release and a commitment to products that have already been announced and work has started on them.
While I dont doubt that Oracle is trying to force customers to its SPARC processors, I feel that Itanium is dead and Oracle is partially right.
It was hp that had the Alliance Agreement with Oracle, not Intel. Do try and keep up.
HP conveniently forget how they left customers high and dry when they dropped Neoview.
They abandon neoview are years of graft to port it to Itanium so they can get together with Oracle to do exadata V1, Oracle buys Sun and ditches the HP kit for the Sun kit for Exadata V2, leaving HP to then flee to the arms of Mickeyshaft for their data warehousing appliance, Frontline or something.
HP has gotten well and truely fecked, its the slag of the IT world, going from the arms of one software vendor to another, and now its got crabs...it's what happens when all you do is push tin..
Why are you not verbose today ?
Reminds me of the old cartoon about a big muscular guy kicking sand at the wimpy kid on the beach. The wimpy kid comes back after weight lifting and kicks the big kids butt. Since HP is just a tin seller and I don't know of any meaningful middleware companies left to acquire I don't see HP ever getting the upper hand.
The most interesting line is how HP complains about how Oracle raised the price to be equal to Power7. "There is no technical or business reason to raise the price". Maybe that is because HP knows Power chips have cores which are 2.5X faster than Itanium cores.
HP/Oracle alliance agreement....I saw nothing in the lawsuit that shows a binding contract to support Itanium.
Can't wait to see the blacked out information about Mark Turd.
Seeing that almost everyone has dropped out of the Itanium $10B alliance will be very interesting to hear about the truth of the Itanium marketing.
Whereas Intel and Chumpaq spent years telling the market that IA64 was going to be the industry standard 64bit processor and that x86-64 wasn't possible, let alone sensible. That worked well for Intel, its system builders and their potential customers, didn't it Matt.
@Todd Rundgren re "Why does HP carry on with the Itanium processor? "
Perfectly fair question, can't understand why anyone would downvote it (at least without an explanatory comment).
All the HP OSes in this picture (HP-UX, NSK, VMS) have already had at least one chip architecture transition and (if HP hadn't demolished the required engineering capability) could relatively easily undergo another one to avoid the unnecessary duplicate expenditure which HP and Intel will in due course stop to allow HP to invest more in more profitable ink instead.
Putting software support to one side, what can IA64 hardware do today that x86-64 can't? Ultra-massive SMP systems, that's all. Opteron and Xeon only go up so far (48 or 64 cores, 512GB or 1TB of memory in today's Proliants) - but that's today, who knows what tomorrow holds.
Actually, we can make a guess what tomorrow holds: AMD64 is very unlikely to go away in the foreseeable future (and may grow more cores and more address bits if the demand exists). Can anyone truly say that about IA64?
Meanwhile, Intel's AMD64 clones now have the same basic system interface as their IA64 chips (memory and QPI). And even Intel VPs are admitting there's no RAS (reliability availability serviceability) reason to choose IA64 over Xeon. And Intel won't stand up in public and defend IA64 in the current slanging match.
HP-UX, NSK, and VMS could run on AMD64 if HP wanted. End customers either wouldn't care about the difference or would be happier than they are with IA64. Perhaps there'd have to be a bit of work done to address the ultimate high end. But that's it. Would that be Utopia, Todd?
First Allison Park was anon coward. And now Matt Bryant has turned to one. What's with the secrecy folks...
Irony, because I am one too! :)
I never post as AC.
"Whereas Intel and Chumpaq spent years telling the market that IA64 was going to be the industry standard 64bit processor..." No-one predicted the AMD64 revolution. At one point, even Sun and IBM were on the Itanium bandwagon. IBM even managed to sell 10,000 IBM Itanium servers DESPITE their sales trying desperately to convert any Itanium opportunity to Power. Sun decided that running Slowaris on the same paltform as AIX, VMS, Tru64, hp-ux, Windows and Linux probably wasn't a good idea as it would merely show up how bad Slowaris compared, so they gambled on UltraSPARC instead. In the end, Slowaris and AIX ports to Itanium were hidden away, which meant hp only had to worry about Compaq's VMS and Tru64. When hp bought them it meant hp was the only company able to offer a real generalised OS for Itanium, so it's hardly surprising that hp soon mopped up 90+% of the Itanium market. If Sun had stayed with Slowaris on Itanium they might have managed to survive, if only by shipping their OS for hp Integrity servers.
".....if HP hadn't demolished the required engineering capability...." FUD. There's still plenty of software and hardware engineering capability in hp, or had you failed to notice that it's the largest IT company in the World?
"......even Intel VPs are admitting there's no RAS (reliability availability serviceability) reason to choose IA64...." I assumed you're referring to Kirk Skaugen, from the XEON side of Intel. Hardly surprising that he's talk up the new E7 Xeons given that the Xeon and Itanium sides of the business are completely seperate and siloed, and Skaugen wants some more of the enterprise market than he's currently getting from ex-Sun shops migrating off SPARC. Not surprisingly, TPM also refers to Skaugen's comments in his report, spendign the whole articel implying the new E7s are only a threat to Itanium and avoiding the fact that they pose a big threat to IBM's Power too.
"......HP-UX, NSK, and VMS could run on AMD64 if HP wanted...." That would be a massive porting exercise. One of the big advantages of Itanium was it was designed as a porting platform, it has features like equal-endianess, which are unique in the industry. Itanium also has a massive more number of registers than Xeon, which means it can simply do more with each cycle. Trying to go down from there to Xeon would be a far harder task than going from Alpha or PA-RISC to Itanium. I have seen a quad-socket Integrity with dual-core Itaniums trounce a quad-socket server with quad-core AMD64 CPUs, with both Oracle on Linux and M$ SQL on Windows. Trying to pretend that just because Xeon and Itanium share QPI and a few other features makes them equivalent is just ignorant.
"features like equal-endianess, which are unique in the industry."
Two I know from personal experience: Alpha was endian neutral although for fairly obvious reasons it only ever sold in DEC-compatible mode; ARM is endian neutral and *is* available in different implementations.
Wikifiddlers assert that MIPS and PowerPC and later SPARCs also have adjustable endianness.
Not a good example of IA64 uniqueness. Feel free to have another go, just in case anyone else is still reading.
"That would be a massive porting exercise. "
It's been done before. Doing something the second time is usually quicker and cheaper, unless the knowledge from the first attempt no longer exists. In HP's case, what remains of the OS engineering and support capability that existed five or ten or more years ago? Does the dif
"HP is the largest IT company in the World?"
Does it feel like that, to folks committed to HP-UX, NSK, VMS? How does it feel compared to ten years ago?
".....Alpha was endian neutral...." Alpha doesn't exist anymore so isn't "in the industry" anymore. Do try and keep up with developments. Besides, IIRC, Alpha was little-endian but had a compiler trick that allowed it to TRANSLATE big-endian instructions, but was not natively equal-endian.
"......ARM is endian neutral....." ARM is at best a micro-server CPU design, definately not an enterprise CPU. More fail, please try again, but please get an adult to help you.
"....It's been done before...." No, it hasn't. Going from Alpha to Integrity was a step up in features such as numbers of register, so it was more a recompile than a rewrite. Going to Xeon would be like trying to squeeze a gallon into a pint glass. Your only option would be a complete rewrite, to go the time-honoured Intel route of breaking the instructions up into smaller chunks and doing in twelve-odd cycles what an Itanium can do in one. Even with more cores working in parallel, the Xeon would still be slower.
".....Does it feel like that, to folks committed to HP-UX..." <Yawn> FUD-level 10! Seeing as hp has a longer and more detailed roadmaps for their OSs and Intel has a longer roadmap for Itanium than IBM has for either AIX or Power, I'd have to say I'm a lot more comfortable with hp than IBM. Don't mention Snoreacle as I'd only laugh. Try again, little troll.
You're a laugh a minute Matt.
According to one of my colleagues, almost exactly ten years ago, Alpha was ten years into a predicted and roadmapped 25 year lifecycle.
Two weeks later, Alpha was dead.
In the UK, the man who had for years been telling customers that Alpha's future was secure and then two weeks later was telling customers that IA64 was the way forward was one Richard George, Alpha Product Manager. Obviously there'd be equivalents in other countries doing similar jobs.
Back to 2011. What do you think Intel UK's Enterprise Channel Director is telling customers about the future of IA64 in the enterprise, Matt? That director's name, by coincidence, is Richard George. Do you think they might be related? I think we should be told.
Roadmaps are great, aren't they, readers.
"IIRC, Alpha was little-endian"
Trust me, you don't recall correctly. The systems Alphas sold in and the software that ran on Alpha made it DEC-order (x86 order) rather than M68K order. The underlying Alpha architecture didn't care. Ditto ARM. Whether you like it or not, this feature is not and has never been unique to IA64.
"doing in twelve-odd cycles what an Itanium can do in one."
Hey, your Xeon mates at Intel tell us that mutli-cores can work miracles every clock cycle, and that even without multi-core, superscalar single-core Xeons can do multiple instructions per clock cycle, courtesy of all that cache on the chip. Are you saying it's not true, or is it just that your ignorance is showing?
"Going to Xeon would be like trying to squeeze a gallon into a pint glass. Your only option would be a complete rewrite,"
Sensible readers will realise that you really mean that the OS sources would need to be recompiled for the new target. Most of it would be a simple recompile, like it was when porting Windows to IA64 (while it lasted) from x86-64. Obviously there'd be a few chip-specific modules (typically memory management and the like) needing detailed attention but the vast majority remains unchanged regardless of The Chip Inside(tm). Compilers are quite capable of handling stuff like different numbers of registers; its what compilers do.
Most of this porting is not rocket science, and you've assured us that HP still have the OS engineering capability, so what could possibly go wrong?
Anyway, shall we go back to waiting for the court case before you embarrass your employers any further?
I was just going through some old posts of mine from 2008 and what do I see, a reply from Matt talking the same drivel 3 years later! You need to get out more mate. The resident HP Troll/Apologist.
As for the old Alpha / Itanic debacle. As I said in 2008, having been promised all sorts of stuff verbally and in NDA roadmaps from HP, AdvFS, TruCluster were being ported. The DL785 was round the corner. All utter sales crap - I agree it was small print issue.
Anyway we dumped our Integrity servers, bought Sun's equivalent of the DL785 in 2006 and got much more ooomph, reliablity than Itanic and H-POX for a fraction of the price.
Saying that H-POX has caught up in a lot of features since those days. However the support matrix is still utterly useless compared to everthing else.
It's a dying platform, only chosen by those already so far down the HP route the pain would be too great to move - though they'll have to one of these days as we did with the demise of Alpha.
Not that I'm a fan of Oracle and what they've done mind, shameless commericial knife in the back. Though what they report Intel as saying is most likely true given the E7 and those roadmaps and no matter how much HP protest and Intel mutter no-one really beleives them.
If I was running H-POX, I'd be looking for an exit strategy asap.
Hey Mr. Sunsetter, HP is turning into Dell, and you know it. HP will not be able to compete with Dell. This is why they brought on Leo, 'cuz they need software. Of course, Sun was trying to do the same thing when their market was shrinking... luckily for them they found the SW Nirvana when they were acquired by Oracle.
Unless HP gets bought out or merges with SAP, HP will just be another Dell.
"You're a laugh a minute Matt...." I would guess there's a lot of laughter in your life, just mostly at you rather than with you.
".....Richard George, Alpha Product Manager...." Yes, I've met Richard George, and, TBH, thought it was a bit of a tw*t. But he didn't make Alpha strategy, just repeated it in the UK. Similarly, he doesn't make Itanium/Integrity strategy, just repeats it. I'm guessing you were just too guillible to think "Hmmm, maybe this guy is just a clown rather than a ringmaster"?
"....your Xeon mates at Intel..." True, I do have several "mates" from the Xeon side of Intel, and they do feed me info on Xeon. They've also been telling me for years, with each new generation of Xeon, that "this is the one to crush Itanium", and each time they are wrong. It's actually quite shocking how competitive the two different Intel product teams are, especially the Xeon boys, but then they had to endure years of being told Xeon wasn't a "real enterprise CPU" by the Sunshiners and IBM fanbois. But every now and again, I have to bench the products, and I'll run up some comparisons on our UAT Superdome and watch it thrash even the latest Xeons. I'm quite looking forward to trying the new E7s but I'm not expecting a Xeon win. You see, unlike you, I deal with real kit in real environments, and don't just go on whatever's on Wikipaedia.
"....Sensible readers will realise that you really mean that the OS sources would need to be recompiled for the new target....." Really? So, that worked really well for Sun Slowaris, didn't it. I mean, it's not like Slowaris on x64 is even slower than Slowaris on UltraSPARC (and that's SLOW!). Oh, but then it is! Because Slowaris x64, even after being recompiled and tweaked to run on x64 (to the point where application binaries are completely incompatible), was still designed for RISC CPUs. Even with optimisations (that's rewrites, not just recompiles), I'm betting hp-ux on x64 wouldn't match hp-ux on IA64. How can I say that? Easy - because (obviously, unlike you) I have put Red Hat on IA64 up against Red Hat on x64, and the Itanium version was faster. Please, do try and argue it wasn't a level playfield or whatever gumph makes you feel better, but the fact is Tukwila is a better chip for enterprise applications than even the Xeon E7s, and Poulsen and Kittson will maintain the Itanium advantage.
"I was just going through some old posts of mine from 2008...." What's the betting it was a thread where the Sunshiners were repeating the same old FUD, that Itanium was dead, that CMT was going to replace EVERYTHING, etc, etc. You Sunshiners have been saying Itanium was dead for every generation for the last decade! And, in the meantime, we had Rock die, CMT fail to grow out of the webserving niche, and the Sun set in a magnificent blaze of failure. You guys just keep repeating the same laughable material! Well, that is some of you guys, the other Sunshiners have dried up and moved on. So don't be surprised if I use the same replies to show you how wrong you are.
"...bought Sun's equivalent of the DL785..." There really is a sucker born every minute! At least you didn't buy the Sun blades, which all seemed to last a single generation before being killed off, ignored for a year and then resurrected as another market-trailling failure. The Galaxies were equally poor.
"....It's a dying platform...." But it outsells all new SPARC combined in the enterprise arena, so maybe you're suggesting Larry ought to stop supporting Oracle software on SPARC?
"....HP is turning into Dell..." Actually, Dell is trying to do what Sun failed to do, and turn into hp. You see, what killed Sun was they failed to diversify well, so when they got caught out trailling the market and selling rubbish, they died. Dell is trying to expand their product offering and be more like hp and IBM. Problem for Dell is hp and IBM keep outbidding them on the good products that Larry wished he had and the Sunshiners wished they could have had (like 3PAR).
"....This is why they brought on Leo, 'cuz they need software...." I guess you were just too blinded by the Sunshine to realise hp is actually already a big software business (like hp Openview, for example, which was the default choice for managing Sun environments for years seeing as Sun had SFA management tools). Did you ever stop to think that Larry brought in Hurd because Oracle have SFA experience with hardware?
"....luckily for them they found the SW Nirvana when they were acquired by Oracle...." Don't you mean "when Sun curled up and died, and Oracle begrudgingly took on the hardware trash when they only wanted some of the Sun software bits"?
"......Unless HP gets bought out or merges with SAP, HP will just be another Dell." Take a look at the figures, denier. Snoreacle couldn't afford to buy hp, because hp is the largest IT company, fullstop. Even IBM couldn't afford to buy hp. On the other hand, hp or IBM could afford to buy a lot more software companies than Snoreacle can hardware ones. Carry on dreaming. Oh, and BTW, isn't Dell's server numbers doing just a lot better than Snoreacle's?
Biting the hand that feeds IT © 1998–2017