HP: last Itanium man standing Make no mistake: If Hewlett-Packard had not coerced chip maker Intel into making Itanium into something it never should have been, the point we have come to in the history of the server business would have got here a hell of a lot sooner than it has. But the flip side is that a whole slew of chip innovation outside of Intel …
HP did the idiotic decision of killing PA-RISC. I'll never forgive them for that.
As much as I loathe Intel, I would like to see the Itanium fare better, if only to get RISC stuff to take over the x86/64 market. But that doesn't seem to be happening anytime soon.
Maybe HP should revive PA-RISC, or even try to chime in with SPARC. Bring back the real processors!
As I remember it, the HP/Intel tie-up looked like a good fit, even though in retrospect I would say that Intel took HP for a ride.
At the time HP decided to jump to a joint-developed chip with Intel, it was engaged in an arms-race with IBM. Tim said that it was the Itanium that made IBM invest in Power, but in actual fact this investment pre-dated the Itanium by at least 5 years. IBM bumped development of Power, PowerPC, RS64, and Power2-7 at various times, but it has been an almost continuous process if we overlook the stumble that happened with the 64 bit PowerPC 620 processor.
The original RIOS based IBM POWER systems, the RISC System/6000, was launched in 1990, and had been under development for at least 5 years before that. The driver was to be an industry leader in the Open Systems market place, as IBM had at last recognized that there was money to be made.
When first announced, the RS/6000 model 530 killed everything on the market stone dead, it was so much faster. HP had PA-RISC running in their MPE/iX line at the time, but it was not a single microprocessor, being built from discrete logic. The RS/6000 caused a huge stir, both because it was so much faster, and also because IBM put significant marketing weight behind their new systems. Sun were immediately knocked off the top of the workstation market, and never managed to really get back up there, and DEC invested heavily to try to produce a really hot chip in the Alpha, that was as a result of the need for speed, was significantly flawed and never really delivered on it's promise.
HP rushed systems to counter the RS/6000 based on the single chip implementation of PA-RISC, and running HP/UX. These were the HP 9000 model 720, 730 and 750, and the race was then on between IBM and HP to see who could have the fastest system. This reached it's peak in the late 1990's, when some models of RS/6000 had marketing lives of less than 6 months.
This was tremendously expensive, and HP, who did not have a big chip-fabrication division valiantly struggled to keep up, but was ultimately doomed to fail.
The way I remember the Itanium being pitched was that Intel were going to take on the development of the PA-RISC single microprocessor replacement, keeping most of the instruction set, but putting in features that would allow the processor to also run x86 binaries, and enhancing the x86 architecture for 64 bit. Intel would get access to HP's IP for the PA-RISC (which included high clock rate silicon and cache IP), and would use their considerable chip making skill to drive the product forward. HP would get a class-leading processor to keep their workstations and servers going. At least that is what was said by Intel.
What actually appeared to happen was that they designed Itanium to be their own processor, with less emphasis on making it a PA-RISC replacement, and more on trying to make it an upgrade path for 32bit x86 servers. They delivered it late, and the product did not live up to their claims as either a PA-RISC replacement, or a 64 bit x86 migration path. Intel attempted to use some of the IP to produce high speed x86 processors, but botched it with the Pentium 4, which was ultimately a dead-end.
Because of the delay, the world in general, and HP in particular, started looking elsewhere. HP appeared to loose interest in the UNIX market place, allowing both their own products and the subsumed products from DEC/Compaq (and to a lesser extent, Tandem) to fall into the legacy category. They produced Itanium based servers, but they were never up there with IBM, except in the very-large system market. Only customer pressure has kept many of the OS's alive.
In the meantime, IBM has been left with the only non-Intel/AMD UNIX offering that was actively being developed, and as a result, has kept market share. Even though there has been no real competitive pressure, IBM has used the convergence of the AS/400 and RS/6000 lines, and to a lesser but significant extent the z series, to move the architecture forward. They have borrowed from other IBM systems (and their competitors) to introduce type 1 hypervisors, hosted application partitions, and a pretty much unrivaled virtualization capabilities. The supported filesystems have scaled, the support for other technologies such as SAN and SVN has gone hand-in-hand with other IBM products.
... it reminds me of ancient history:
http://www.zdnet.co.uk/news/processors/2001/10/16/arm-to-unveil-64-bit-jaguar-chip-for-handhelds-2097352/
(sorry for cross-linking to non-reg sites)
ARM is great for what they are. To play in the datacenter with the really big guys, we're talking 64bit, terabytes of RAM, dozens of I/O busses. Almost ten year old press releases notwithstanding, ARM hasn't delivered any of that yet.
First of all the glowing forecasts for Itanic were based on delivery forecasts for the different iterations in its development. Itanic was promised along with 4GHz+ Pentium 4s, NPUs, StrongArms and a long slew of other chippery by the Intel R&D division at the time. The whole division however failed to deliver in the most miserable way. Apparently, they were more interested in fiddling expenses than doing work. We all know where that division was located.
What saved Intel was a set of skunkworks in another country (Israel) which broke the party ranks and used the old P3/PentiumM for development of what wee now know as Core.
Giving Intel credit where credit is due, once the level of non-delivery and fraud in their original R&D center was discovered they closed it (instead of doing the classic "UK manager" thing and joining the embezzlement bonanza).
As a result however they had a dead NPU business, dead StrongArm business and dead Itanic business. The alternative R&D location in Israel had resources to accommodate only x86.
That is all this is about. It is a tale of a miserable failure in "moving R&D to a country with lower costs". If Intel kept the Itanic R&D in the USA it is quite likely that it would have been an Itanic machine to process this post today.
All your arguments here hold equally true for Power - a short term perf advantage isn't going to rescue IBM's Microelectronic business from its ongoing loss making position (unless the US Govt decides to pay IBM to develop P8, the way it did P7). And arguing that Power still has RHEL is a nonsense as _everyone_ knows that Linux on anything other than x86 is and always was a boutique solution.
What you miss here of course (and given your strong IBM leanings it doesn't really surprise me), is that last time I opened up a server and looked inside there was MORE IN THERE THAN JUST A PROCESSOR! Innovation doesn't start and end with processor designs. Both IBM Power and HP Integrity systems are exactly that - "Systems".
Also have you considered that if HP doesn't appear to deliver many innovations in the HP-UX space its cos it's customers want a slower pace of change to deliver longer systems lifecycles? Ask any IBM Power shop what they find most frustrating about Power, and I'll bet more than half will tell you it's the short support lifecycles of AIX - customers in this space value longevity more than sparkly new features they may never use (just go ask your buddies in the IBM System Z division)
"and I'll bet more than half will tell you it's the short support lifecycles of AIX.."
You've gotta be kidding. It's always a balance. You want you OS to move forward but not so fast that you can't keep up. Aix versions have had a lifespan of around 10-12 years, that is IMHO ok.
And funny enough that is exactly the same lifespan as HPUX'
http://h20338.www2.hp.com/hpux11i/downloads/public_hp-ux_systems_support.pdf?jumpid=go/hpuxservermatrix
But IMHO AIX is better at versioning.
I mean the number of times in my last job I have had to deal with stupid project managers and architects that panicked, cause their software stack didn't work. cause "the support matrix said HPUX 11i ", and you've had to dig through manuals and foras just to find out that the supported versions were 11v1 and and...
And while you might complain that AIX is evolving to fast, then yes it is IMHO going a bit perhaps to fast, but that also due to the hardware and virtualization.
I mean back in 2H 2001 a 2 socket Dell Itanium workstationwith a 800 Mhz Itanium processor had a base score of 7.15 on specint2000_rate versus 7,2 for a 2 socket POWER3 450 Mhz POWER3 IBM workstation. Now that was pretty close, back then. Today the brand new Tukwila will do 134 specint_rate2006 on 2 sockets and POWER7 will do 652 also on 2 sockets. Now that is a factor of almost 5 on 9 years that POWER has take a lead with.. so..
Where I work we are slowly starting to move from 5.3 -> 6.1
But I still once in a while struggle with customers in who are running 4.3 or 5.1, and then it's migration time.
// jesper
The article claims that the threat of Itanium caused increased innovation in its competitors, specifically IBM's Power, AMD64 and Sun's Sparc, and calls that the real success of Itanium.
However, the article forgets that Itanium killed several competitors long before it was even out on the market: Compaq bet on the Itanium and stopped further development on the Alpha CPU that it "inherited" from Digital and SGI likewise stopped development of its MIPS server processors. HP also dropped its own PA-RISC, but by having a more active role in the Itanium project, they had better reasons than the others.
So, rather than promoting innovation in its competitors, the Itanium killed several of them. In the end, the real winner in the server market was the x86, so it all worked out for Intel's good.
The strategy was:
Embrace AMD to force HP's hand on offering AMD system
- Intel will release 64-bit Xeon as a response to HP offering AMD (check the dates it happened)
- Force Intel to retrench in Xeon vs. Itanium to protect their "core" business
- Invest heavily in Power to dominate the Unix market
....And Rick Belluzzo did a great job running SGI into the ground after he forced them to dump MIPS and convert to Itanic
Thanks RICK!
There's not doubt that the x86 architecture can, or will be shortly able to handle 99% of what Itanium will do. However, the big problem that many of us is migration - especially of huge multi TB databases (those of us with databases in the several tens of TB have a problem). The normal methods of migrating such databases with minimal downtime (creating a copying and synching using physical logs) won't work with binary incompatible databases. There are logical replication tools, but they are expensive, bespoke and often have performance problems. That's without issues of binary compatibility of flat files.
As x86 is little endian and HP-UX on Itanium is big endian, then this is a huge issue as many of us have windows for migration measured in a few hours (at most).
"last time I opened up a server and looked inside there was MORE IN THERE THAN JUST A PROCESSOR! "
Well, like, yeah!
But so what?
Intel's Hypertransport knockoff is now used on both Xeon and Itanium ranges, and the memory technology is fundamentally the same too.
So, what's the difference between a "system" based on IA64 and a system based on AMD64?
Answer: cosmetics, vendor politics, and, in the bigger picture, the OS. Quickpath and common memory technologies mean that there are no fundamental technology differences between AMD64 and IA64 systems, and sooner or later the sole remaining IA64 system builder will have to acknowledge that.
You could argue (they certainly will) that there is still a difference in ultra high end scalability and that IA64 is the only platform for huge-SMP huge-memory systems, but when AMD64 Proliants are already up to 48 processors and (half) a terabyte of memory, does much of the market really see that as a limitation? And who knows where AMD64 will be in 12 months time...
--"And wouldn't it be far easier if HP just bought Red Hat and asked its HP-UX customers to make just one more port and compile, to RHEL running on x64 iron?"--
No no no no NO, get this silly notion out of your head. That would be a disaster for Red Hat: they need to stay vendor neutral in order to deploy on as wide a variety of vendor hardware as possible.
And anyway, why would HP spend billions buying Red Hat to get a distro of their own when they can copy what CentOS and Oracle have done? They can have their own HP branded Linux any time at a fraction of the cost and share the development going forward with Red Hat and Oracle.
--"Novell would be cheaper to acquire and still supports Itanium with its SLES Linux, and that would also do"--
No disrespect to the guys at Novell, but they have all the tools, IP and products to really make it in this business. If they can't make it work well given what they have then maybe they do need a change of management. I can't see HP going for them though as the bulk of their linux customers want Red Hat.
It's a bit like the old Kim il Jong stories about how North Korea is actually the most prosperous country in the World and the rest are just puppets of a decaying United States, out to destroy the glourious North Korean People's Republic! TPM and co (especially Peter Gathercole) seem to have very selective memories.
For a start, trying to pretend IBM Power has had some mythical lead over the other commercial UNIX CPUs such as SPARC and PA-RISC? That everyone was just dying to port their software to Power first and then SPARC and PA-RISC as an afterthough? Laughable! A simple example would be the Veritas suite of software - how many years did IBM AIX lag Slowaris and hp-ux on Veritas? Which kept Power well out of the high-end business critical roles where us customers wanted a common stack (with Veritas) that could be run on two flavours of UNIX (yes, hp-ux and Slowaris, not AIX), so we could play one UNIX vendor off against another. IBM simply wasn't considered. And don't even try and compare IBM's lukewarm support for Oracle (after all, IBM wanted to sell DB2) compared to Sun's or hp's efforts, which was another reason not to look at AIX-Power.
IBM and Sun backed out of the Itanium party because they realised they would not be able to compete as effectively if it was on the same chip as everyone else, especially as hp had such an inside on the chip design. Sun thought they could bank on their massive installed base to out-last the other commercial UNIX vendors. In IBM's case the idea of a common UNIX platform was a massive threat to the mainframe business that IBM has used to prop up Power for years. AIX was far too weak to go up against either Slowaris or hp-ux, especially in mission critical areas. HACMP and it's lag behind tools like ServiceGuard or Veritas Cluster is a simple example of where IBM kept AIX "dumbed down" so it didn't steal sales from their mainframes. IBM rightly guessed that they could fleece their mainframe customers to keep their UNIX business going, and it is only the mainframe business that has kept Power viable (shame that didn't work for Cell). But with mainframes declining and the UNX pie getting smaller every year there is only a finite number of years IBM can milk those mainframe customers. With them being far from dominant in x64 (and having abandoned desktops), IBM risks becoming a second tier vendor when x64 finally does finish eating the UNIX market.
And let's get past this myth that it was only customers "stranded" on any OS from one of the other Itanium partners that were forced to buy Itanium when they really wanted Power. For a start, IBM was at one point the second largest Itanium seller. IBM sold 10,000+ Itanium servers despite trying as hard as possible NOT to. IBM's own sales teams found that many deals where AIX and Power just couldn't do the job. Please try and pretend that was 10,000+ development systems, just for a laugh.
Then you whitter on about how Intel developed the original Itanium in their image, when the fact is hp had already developed the core that became the Merced release of Itanium long before approaching Intel. Subsequent developments by Intel have developed that core far beyond the original hp design, but the core is still based on that Merced design.
Every Power generation we get the same hot air and frothing from the IBMers - "Power is great, Power i sooooo fast, no-one will buy anything other than Power!" And every version - P4, P5/5+, P6/6+ - the market goes out and buys other vendors' systems because they realise the system is not the CPU alone, and that IBM does not offer the best solution for all roles. I know the once Sun-centric Reg has taken a massive lean to the IBM camp since Ashlee left, but you guys want to watch that lean, it's looking more like you're falling flat on your faces.
/SP&L
Start with a cage containing five monkeys.
Inside the cage, hang a banana on a string and place a set of stairs under it. Before long, a monkey will go to the stairs and start to climb towards the banana. As soon as he touches the stairs, spray all of the other monkeys with cold water.
After a while, another monkey makes an attempt with the same result - all the other monkeys are sprayed with cold water. Pretty soon, when another monkey tries to climb the stairs, the other monkeys will try to prevent it.
Now, put away the cold water. Remove one monkey from the cage and replace it with a new one. The new monkey sees the banana and wants to climb the stairs. To his surprise and horror, all of the other monkeys attack him.
After another attempt and attack, he knows that if he tries to climb the stairs, he will be assaulted.
Next, remove another of the original five monkeys and replace it with a new one. The newcomer goes to the stairs and is attacked. The previous newcomer takes part in the punishment with enthusiasm! Likewise, replace a third original monkey with a new one, then a fourth, then the fifth. Every time the newest monkey takes to the stairs, he is attacked.
Most of the monkeys that are beating him have no idea why they were not permitted to climb the stairs or why they are participating in the beating of the newest monkey.
After replacing all the original monkeys, none of the remaining monkeys have ever been sprayed with cold water. Nevertheless, no monkey ever again approaches the stairs to try for the banana. Why not? Because as far as they know that's the way it's always been done round here.
And that, my friends, is why Matt Bryant still stands up for Itanic
Careful, Matt, or you too will be falling somewhere inconvenient.
Most of the stuff you talk about as differentiating HP's IA64 offerings from IBM's high-end or mid-range offerings (or Sun's offerings) is ***pure software*** (Veritas Cluster etc). It's only on IA64 for historic reasons, presumably because HP chose to pay to have it there.
For example, Veritas Cluster already runs on Compaq/HP Proliant servers (x86, not IA64) and there were (are?) even HP storage packages based around it.
If the only differentiator between x86 and IA64 is not hardware but is OS-class software, surely it doesn't bode well for The IA64 Inside in the medium term? Is there any other differentiator?
".....Most of the stuff you talk about as differentiating HP's IA64 offerings from IBM's high-end or mid-range offerings (or Sun's offerings) is ***pure software*** (Veritas Cluster etc)...." Yes, which is exactly the point! If the AIX-Power combo was so much amazingly better than anything else on the market then history would show a rush of software vendors working on AIX first, then Slowaris (because of the installed base), then hp-ux as an afterthought. However, the last ten years have seen software vendors working much harder on Slowaris and hp-ux software than AIX, with Veritas being a very good example. It wasn't until late 2002 IIRC that AIX finally got the same tools as Slowaris and hp-ux had had for years, despite IBM paying Veritas to start a joint dev program back in 2000. That joint dev program came around when IBM finally admitted it was obvious that having common Veritas toolsets on Slowaris and hp-ux was allowing Sun and hp to gain more of the enterprise UNIX pie.
The Oracle example is equally valid - over the same last ten years, IBM tried to push DB2 on AIX on Power, which didn't compare to Oracle on either Slowaris-SPARC or hp-ux-Itanium. Oracle's own customers pushed for Oracle on hp-ux to be given equal development speed as Oracle on Slowaris (Sun was Oracle's prefered partner for years). There wasn't any similar push for Oracle on AIX-Power. All of which seems to blow great big holes in the amusing idea that customers rushed to AIX-Power in preference over hpux-Itanium or SPARC-Slowaris.
".....If the only differentiator between x86 and IA64 is not hardware but is OS-class software, surely it doesn't bode well for The IA64 Inside in the medium term? Is there any other differentiator...." Scale. And the available apps that take advantage of that scale. Outside that, there is very little reason to run anything on commercial UNIX on a trad RISC/EPIC platform when you can run it on Linux on x64. But, seeing as hp dominate the Linux on x64 market, I can't see them being too upset about that!
At least Matt is only revising past history and attacking Power instead of lamely trying to defend Itanic (notice he speaks of its glory in past tense?). I guess even the fanbois have admitted defeat. Like I say the reason Intel wants to kill off Itanic is because it is a painful reminder of their hubris ala the Cadillac Cimarron - http://en.wikipedia.org/wiki/Cadillac_Cimarron - (lol wonder if they also have a Itanic chip design picture hanging in the CEOs office saying "Lest we forget"). All in all Intel played HP sick for their IP but HP is ok with that because after all you don't need Itanics to sell printer ink. Hell HP might long outlast even Intel (technologies come and go but people love printing shit and have for over 300+ years).
Spot the Sunshiner - still can't get over the hp ink bizz making more money in a quarter than SPARC ever made Sun! It's amazing that they still can't get a hold of that whole product diversity idea, as though having a product outside of UNIX that makes a profit is somehow beneath them. Don't worry, your defeat has already been widely acknowledged, or did you just forget that whole Sunset episode?
If any product should be refered to the Cimarron then I would suggest the Rock is a far better candidate. Then again, that would be unfair to the Cimarron - at least the Cimarron made it into production!
/SP&L @ the Sunshiners, the gift that just keep on giving!
HP makes a sh*tload of money on INK. That is why HP is normally referred to as BIG Ink :)=
And no I don't buy HP ink cartridges, I use buy refurbished ones as I have a HP scanner, HP Printer, HP scientific calculator....
Well who cares anyway, there is football on the TV and beer in the fridge.
// Jesper
So is the VLIW/EPIC dream of increased instructions/cycle over RISC/CISC designs dead, or did Intel just botch the Itanium design so badly that it can never deliver? In any event, moving to multiple-core chips may be the path of least resistance for now, but what happened to the talk about 20+ IPC per core from their Elbrus acquisitions?
"So is the VLIW/EPIC dream of increased instructions/cycle over RISC/CISC designs dead, or did Intel just botch the Itanium design so badly that it can never deliver?"
Mostly the latter. The idea of VLIW is to keep the hardware simpler than OOO processors, which potentially increases clock rate. Intel added lots of features to this concept ending with a hugely complex processor, thereby defeating the premise. Merced had a lower clock rate than contemporary x86s and much lower than Alpha, which was designed to be simple. Intel also relied on compilers being able to exploit the complex hardware, which didn't happen overnight. So, apart from a few benchmark programs, the potential of the hardware was never really achieved.
However, the single-processor advantage window of VLIW was fairly short: In the 90s, increases in clock rate was the main driver of increased performance, but in the 00s, clock rate hit a ceiling: Increasing it further made the chips use too much power and require more cooling. Hence, the last 5 years have not seen any significant increase in clock rate, and the development has turned towards multicore processors instead. So the fact that VLIW allows higher clock rate is of less importance now. A simpler processor may allow more cores per chip, so VLIW may still have a future, but I think mostly in scientific computing. Most other types of programs today are too dynamic for VLIW, which relies on static analysis to get ILP.
Itanium's architecture with the EPIC instruction "bundles" - simplified, putting a signal processor's programming model under a general-purpose instruction set - would fly on paper but often fall foul of real-world issues. Namely, how do you feed the parallel pipes - how do you keep interdependencies out of the instruction bundles ?
Intel's answer was "the compiler / toolchain / feedback optimization will figure that out for you".
Around 1999/2000 when you could get the "merced simulator" from HP and look into the actual code generated by the then-available compilers (ms c, intel c, sgi, gcc), it was very noticeable that much of what was generated was "padded bundles" - VLIW instruction words with a single effective instruction, and all other slots nops. Intel's compiler was quite a bit better exploiting the parallelism but still 50% of the instruction slots ended up as nops.
So real-world code transformed your wonderful parallel EPIC into wide-but-shallow instructions eating up memory bandwidth.
Thing is, if all you use 20+ IPCs per core for is to do more of nothing, will your programs be faster ? Will you win a F1 race in a 10000hp lorry ?
Matt, Veritas softwares were always bundled together with Solaris until ZFS came out.
AIX didnt need any of tools because of the JFS and then JFS2 free of charge.
And, dont tell us PowerHA (not HACMP for sometime) is lagging behind HP MC Service Guard.
This is real FUD.
Give us an example what are the premium features that MC has that PowerHA does not.
You are just a clown! Shame on you!
Cheers!
"Matt, Veritas softwares were always bundled together with Solaris until ZFS came out...." You missed my point - Veritas released products on Slowaris and hp-ux because that's what they're customers wanted. IBM's AIX lagged years behind either. I agree that Sun switched to ZFS though probably to save on paying Veritas licensing costs.
"....AIX didnt need any of tools because of the JFS and then JFS2 free of charge....." Having tried to get IBM AIX kit working with the Veritas tools we already used on Slowaris and hp-ux, I'd laugh at your comment but the memory is too painful. I'd rather use Veritas on Slowaris than AIX's versions, though don't tell the Sunshiners that!
"....And, dont tell us PowerHA (not HACMP for sometime) is lagging behind HP MC Service Guard....." We're talking history here (though I personally think PowerHA still lags MC/SG, but then that's just my opinion). Historically, HACMP was a complete dog compared to ServiceGuard, and the problems upgrading between version were legion! It wouldn't have been so bad if AIX's option for partitioning hadn't always lagged hp-ux as well - how many years after npars was it before AIX got proper hardware partitions which were actually electrically isolated? Have they even got them yet, even with P7? And I still remember the fun of IBM trying to sell us their Integrated Virtualization Manager software, which didn't work with their PLM software or their Hardware Management Console, and wasn't even supported on the P570s they were recommending! In comparison, hp's offerings for partitioning and clustering have always been much better integrated, especialy when it came to management tools.
Try again!
/SP&L
Obviously you won't, unless the F1 rules change to say you have to tow a fully loaded 38 tonne trailer. Which could be more fun than today's F1.
And nor will you get round a congested M25 (London Orbital Car Park) much faster in a supercharged Porsche than you will in somebody's 3 cylinder four seater diesel.
IA64 VLIW is a bit like planning the fastest possible route around the M25 for the Porsche, and then realising that in the real world the performance is limited by factors which can completely be foreseen (you *know* there will be congestion) but which cannot be worked around (not least because you don't know exactly where or when the congestion will be). You might as well have bought the three cylinder diesel, unless you have a private racetrack to play with the Porsche on.
Mine's a pint (not when driving, obviously).
... It was the Sun Starfire system (Sparc), which completely reinvigorated the entire Unix market. Without the Internet and Sun buying the Cray Business Division when they did, we'd possibly all be running on Itanic today. Shudder...
Itanium is a huge failure. If Intel is only making less than 2Billion a year, then that can hardly account for the R&D and the "investment" from/to the Itanium Solutions Alliance.
Damn it! Don't make me defend MB.
"Matt, Veritas softwares were always bundled together with Solaris until ZFS came out."
No, Veritas software was never bundled together with Solaris. It was bundled with Sun's early fibre disk solutions (SSA/A5000/etc), but not with Solaris. Sun did bundle Solstice Disk Suite (SDS).with Solaris.
"And, dont tell us PowerHA (not HACMP for sometime) is lagging behind HP MC Service Guard."
PowerHA is just a renamed HACMP. Even IBM says PowerHA (HACMP). Of course it's better now, but still behind the competition.
"Give us an example what are the premium features that MC has that PowerHA does not."
You sound like an IBM salestard. Sell features, but forget that many are in future releases, or actually on other products, or not supported on the system that we'll actually need to use it on. Feature lists are for the Execs, not for actual users. The last time we played with PowerHA/HACMP it required so many custom agents that may or may not have worked we gave up. Not much support for 3rd party apps in other words. Of course GS would have been more than happy to sell us a service to create the agents for us...
"......It was the Sun Starfire system (Sparc), which completely reinvigorated the entire Unix market....." Very debateable seeing as the majority of the SPARC base has always been low-end webservers. Sun backed out of the Itanium alliance long after the Starfire systems arrived because they saw that they could bot differentiate enough on Itanium compared to hp or IBM. That IBM also thought the same of Suna nd hp makes it all the funnier! Instead, Sun sailed on into the Sunset with a reliance on SPARC that slowly killed it.
"....Itanium is a huge failure...." Well, seeing as it's alive and still in production compared to Sun SPARC, which is dead, I'd say that would have to make the whole SPARC saga a complete and total disaster of epic proportions!
"......If Intel is only making less than 2Billion a year, then that can hardly account for the R&D ....." Strange then, that when Sun was sinking fast and had less than $4bn in the bank, the Sunshiners insisted that was more than enough to carry on developing Rock for years! And Intel doesn't have to develop the servers to go with the CPUs as the Itanium Alliance vendors do it for them, so much less expensive R&D for Intel than IBM. Of course, Sun also didn't have a really successful partner chip like Xeon to share development costs with (Niagara is a tiny niche product in comparison), so it's not surprising they just don't get the idea of the advantages that economies of scale bring.
Let's put an end to all the IBMer and Sunshiner FUD and just agree a few facts, shall we? Itanium in Tukwila is alive and well and has a set of very viable products in the form of the new hp blades ready to go, especially when you consider the large installed base of hp blade chassis they can plug straight into. They will find favour with existing hp-ux and OpenVMS users because they will provide a boost in performance and be less of an upgrade hassle than switching to another OS on another platform. They're also pretty cheap, so IBM will have to really drop their pants to stay in the game. Rock is dead, and SPARC64's future is at best uncertain. Power seems to be around for the next few years, but there is a lot less public info on Power8 and any following IBM chip than there already is on Intel's Kitson and Poulsen. In the meantime, Intel will make money hand-over-fist with Nehalem, which will help keep Itanium development ticking along nicely. You can argue that Niagara has a future if you wish but most people in the industry I know think Nehalem and Magny Cours have made Niagara irrellevant overnight. So, it still looks like a two-horse race in the commercial UNIX space - Power vs Itanium - for a few years yet.
/SP&L
"less of an upgrade hassle than switching to another OS on another platform."
Obviously.
But how about switching to *the same* OS on a slightly different platform? HP-UX on Proliant on x86-64? Or VMS on Proliant? Or the NonStop thing on Proliant? The world's businesses already run on Proliant. They're known to work.
That way, customers get the same wonderful OS, the same wonderful features. That's what the customers care about. Customers (be they end users or systems houses or whatever) don't much care about The Chip Inside (tm), unless it gets in the way. IA64 gets in the way, because it's not x86-64, and as such is in a semi-permanent state of vulnerability.
You know it makes sense. They do too.
Only internal face-saving/politics is stopping this, and sooner or later, money will outvote the internal wrangles.
"....But how about switching to *the same* OS on a slightly different platform?...." If you stop to consider that you see the issues the Slowaris crowd are now facing. Even Slowaris on x64 is not an easy migration from SPARC Slowaris as the binaries are different (different endian) and some Slowaris software vendors have not ported their apps to x64 or are only porting to Linux on x64. And then some of those apps don't scale as well on x64 as they do already on SPARC. Except for those users that have to move for performance reasons, I expect many will simply sit it out on old SPARC for as long as they can, to the point where Soreacle makes ongoing support costs higher than the cost of new kit plus a migration. I predict AIX, hp-ux, mainframes and OpenVMS will follow similar paths.
Then you have to wonder why go to the hassle of migrating everything from one version of the OS on platform A to platform B if there is an already proven solution on platform B that can be bought off the shelf with a different OS? For Slowaris users that means Windows and Linux, both of which have better app coverage than Slowaris on x64. If you look at it that way, why would you bother trying to fight the tide and port hp-ux or OpenVMS to x64? Whilst you could claim the clustering tech is a differntiator, the reality is the only real differentiators are scale and the apps that make use of it. Whilst app support on x64 lags behind the trad UNIX CPUs on scale there will still be a market for those trad CPU platforms running commercial UNIX.
"...That's what the customers care about..." Actually, what the "customers" care about is meeting a business requirement with the least cost, risk and hassle, in a reliable way that gives them at least parity if not a competitive advantage over other businesses. They don't give two hoots over what's actually running underneath (that job is left to us techies to worry about) as long as it does the job reliably and the vendor is seen as offering good support. If I could find a solution that did the job on pocket calculators better and cheaper than Superdomes then my boss would shell out for the calculators. I have worked with companies that have bought obsolete kit simply because they know it will do the job with the stack they have with minimal issues, and that outweighs the concern about ongoing support or the greater disruption of moving to a new platform. Go stick your head in enough datacenters and you'll find stuff like old PDP11 gear still whirring away, despite it not being vendor supported for many years!
A port of Non-Stop to x64 does make sense as neither Linux or Windows offers a comparable completely fault-tolerant solution, but again hits the scale issue. I can see hp offering an x64 Non-Stop solution as an entry solution, but not displacing Itanium completely for a while yet.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
