RE: Well, where's Matt?

Sorry, I was actually waiting for a Sunshiner with a comment that actually contaned some logical argument before posting a reply, but it looks like I'd be waiting a long time. Besides, Sun is dead so not much fun in pointing out their folly to the Sunshiners, I suppose I'll have to switch to baiting the AIX crowd soon.

Let's deal with the "hp is the only vendor wanting Itanium becasue it is totally reliant on Itanium" bit first. Posters using that "logic" manage to conveniently ignore the simple fact that hp is also the leading x64 vendor and Intel's and AMD's largest x64 customer. If hp just wanted to sell x64 then they'd be happy to let Itanium die and wouldn't be pushing Intel so hard. The vast majority of those Linux and Windows boxen that have been replacing SPARC Slowaris servers for years have been hp ProLiant systems. And now that Sun is dead and their customers faced with migrating to another OS, hp is perfectly placed to plunder those orphan Sun accounts, because, unlike Sun, hp engaged rather than opposed the Linux community and Microsoft. But it is not a one-punch strategy, hp also needs Itanium to squeeze the RISC base from above, including AIX Power, whilst ProLiant eats it from below. Without that pincer move, Slowaris refugees could move off SPARC and upwards to Power or mainframe, but Integrity gives both the performance and pricepoint to offer a superior deal. Throw in the easy transition from Slowaris to hp-ux, especially with the porting programs that hp run, and you see how hp is well set to eat up those old Slowaris shops. And seeing as those Slowaris shops are the targets likely to generate the largest areas of growth for hp-ux, Linux and AIX in the next few years, it is no surprise hp and IBM are gearing up to fight over them.

So, will being stuck with the current Integrity servers for another year hurt hp in that battle? In a way, yes, because customers have been led to expect and have planned for quad-cores. To avoid customer disappointment I expect there to be a few more points off deals to convince some customers to take the current servers instead (well, that's what I'll be beating our hp salesgrunt up with anyway). Will it hurt hp because Integrity will somehow be behind in performance? The answer there is probably not. SPARC hasn't managed to match Itanium for years and is now zombified. Rock, even if it does escape Larry's obsession with rehashing the network computer, is going to be comprehensively outperformed by Nehalem and the current Itaniums, and Niagara is struggling to get out of its webserving niche before the boys at Oracle realise just how limited it is. AIX is still waiting on Power7 as we have now seen Power6+ as such a non-event that IBM didn't even make a noise when they slipped it into their range. IBM's gamble on extreme clock speeds hasn't worked out, it is still the surrounding architecture that keeps the cores spinning that is vital, and hp seems able to do that well enough now to match Power6+. Unless IBM manage to pull Power7 forward and also fit it with a radically better surrounding architecture, Intel have time to tune Tukzilla to meet and beat Power7. Then there are Kitson and Poulsen to come, and what does IBM have planned for after Power7?

So are Nehalem and Istanbul the big threats to current Integrity? Yes, just as Xeon and Opteron have been to RISC for years. But according to all the Sunshiners, quad-core x64 was supposed to have killed Itanium by several years ago, but the reality is Itanium (especially in Integrity and with hp-ux) still does some jobs that x64 can't and still does a lot of jobs better than x64. There are even cases where Itanium with Windows beats Xeon and Windows, especially large M$ SQL instances. And it still will even with eight-core x64 CPUs are available. There are plenty of heavy-threaded apps out there that require the grunt of Itanium for best performance and hp-ux for superior RAS and scaling.

And in a way Nehalem is Itanium's best allie, as they allow hp to use the economies of scale of the ProLiant bizz to benefit Integrity as the two ranges now share more and more parts, much more than i/pSeries and xSeries or M-series and Galaxy (just mentioned the Sun kit for the humour value). And tehn there are the continuing worries about IBM's commitment to x64 - will they sell the xSeries bizz to Lenovo like they did the PC bizz? Every account hp gets ProLiant into makes it easier to push Integrity into as they share the same comprehensive centralised management tools and support, and hp can deliver the whole deal with better storage options. There are even hp Integrity blades and NonStop blades that fit into the ProLiant blade chassis - yes, that hp blade range being the dominant blade vendor offering too. So Nehalem and Istanbul will actually help Itanium as far as hp are concerned.

In fact, if we want to talk about reliance on a single CPU design and any subsequent vulnerability, then IBM should be the target. Sun has demonstarted the folly of relying on one range - UltraSPARC - to keep the revenue rolling in. After all, IBM's revenue stream is much more dependent on Power than hp's is on Itanium. The vast majority, I think it is 60+%, is tied to Power. All those mainframe sales, services and licensing streams are largely tied to Power - what happens if Power7 slips some more? And yes, IBM trolls, Power has been late before, simply moving the deadline is slippage, it's not delivering on time. With the majority of hp's revenue streams related to x64, it is sitting pretty with Nehalem and Istanbul, and Itanium is the bonus that allows hp to own the high-end and squeeze AIX in between and attack upwards at the mainframe base. A slip of another year on Tukzilla isn't a massive problem for hp, but a slip of another year on Power7 could see IBM taking a big hit in revenue. But then I wouldn't expect TPM to highlight that.

/going to be so disappointing replacing the Sunshiners, the AIX boys are so grey and boring, they just don't froth and rant like the Slowaris fanbois. Ah well, SP&L whilst I can!