"The job of defending Itanium..."
"The job of defending Itanium gets relegated to the Itanium Solutions Alliance..."
Shouldn't that read:
"The job of defending Itanium gets relegated to Matt Bryant..."
Hewlett-Packard might have been the acquirer when it merged with Compaq back in May 2002, but when it comes to server operating systems, the Compaq inclusionary philosophy usually prevails even if it does take some time to be turned into action. Like something akin to a decade in the case of Solaris. Yesterday's announcement …
"The job of defending Itanium gets relegated to the Itanium Solutions Alliance..."
Shouldn't that read:
"The job of defending Itanium gets relegated to Matt Bryant..."
"With HP and Intel, the main proponents of Itanium at this point, both being partners with Sun with regards to Solaris, the revival of _Sparc on Solaris_ would seem to be long overdue."
That should be "Solaris on Itanium", I would say.
...if you are into entry level.
Should team up fine with Solaris, which is a kind of "pants-two-numbers-too-big-snowboarding-web-monkey-OS" on its own.
GNU for teh win!
Flames, of course.
Obviously, if there is little demand for the Itanium architecture, Intel can't devote much in the way of resources to it, which leads to little reason to demand it.
The benefit of Itanium is presumably that it in some way allows chips to be produced that are bigger and more powerful than x86 chips. That's a very valuable benefit, but obviously it can't be realized if they're a generation behind.
Intel needs to do something.
I still feel that bringing Itanium goodness to the x86 world in some fashion - thus, making x86 chips more powerful, and letting the x86 subsidize the Itanium market with its volumes - is the way to go.
That would be the area with all the high-margin services and support, then, the real money-makers? IBM's very lucratibe mainframe bizz is built completely on that "niche high-end", a fact a mainframe bunny like TPM should know. But maybe TPM should also know that hp's Integrity range scales from two-way blades to Superdome, and all segments are selling well even if Integrity is most dominant in the high-end. Sun's big problem is it is losing share in that high-margin high-end and so not making the profits it needs to. In fact, Sun can't make a profit.
TPM also forgot to mention that as well as hp-ux and Windows you can buy fully-supported Linux and OpenVMS on hp Integrity, whereas ProLiant is just Windows, Linux and now Slowaris, so Integrity actually runs more fully-supported OSs than Nehalem ProLiants will. Even though I enjoy windng up the VMS dinosaurs we have here, I'd be the first to admit OpenVMS is still a very popular OS amongst financial instituitons, with more of a future than Slowaris on SPARC. Both hp-ux and OpenVMS have much longer public roadmaps than Slowaris. In fact, hp-ux's public roadmap goes further into the future than Slowaris, Windows or AIX, and the reason is because customer demand is still very strong.
Sun's decline is partly poor strategy but more poorly performing products - they lost to Linux because Slowaris on SPARC was just that - slow! Customers could switch to Linux on cheaper x86 and get better performance for the majority of the webserving that was the core of the SPARC business in the dotcom bubble days. Hp had always aimed higher with hp-ux. In the late '90s, whilst Sun was making most SPARC sales from one-socket and two-socket webservers, hp's bestseller was the four-way and six-way K-class hp9000 servers, and these were primarily used for Oracle and like databases. A simple example of this difference in approach is that Sun's Slowaris Cluster didn't arrive until 2000, years after hp's ServiceGuard had been accepted in the market. Whilst ServiceGuard became a key technology for hp in the corporate datacenter, Sun Cluster was derided as a toy and Sun was reliant on Veritas clustering technologies when fighting hp-ux in those high-end, business-critical deals.
Whilst webserving was an easy target for Linux, replacing database clusters has proved a tougher proposition. IBM and hp saw this and allied with Linux to attack the easy target of Sun's webserving base. Sun went on the defensive, allienated the Linux community, and has suffered ever since.
Because hp-ux has always had scaling and performance advantages over ProLiant, hp has wisely not tried to play it against Linux but as complementary offerings - Windows and/or Linux at the front, and hp-ux doing the heavy lifting in the back. Hp has made money from both whilst Sun has lost marketshare with Slowaris, and Sun failed in repeated schizo attempts to woo and then attack the Linux community. Hence Sun's desperate attempt to fight Linux with Slowaris x86, which is very obviously failing. Sun's own Galaxy customers prefer to order Linux on Galaxy at a rate of five-to-one over Slowaris x86. Schwartz is happy to bleat about the millions of free downloads of Slowaris x86 but hates to admit the numbers of downloads that convert into support contracts as they number in only the thousands. And the servcies pull through is simply non-existant.
Sun's Project Copy Linux has failed, and now Schwartz is just looking for a means to keep the core software business going by outsourcing his server hardware business. Hp will be happy to mop up more SPARC customers as they have already gained many without Sun's help, so Schwartz is finally being realistic and not trying to swim against the tide. Sun's desperation to sign up hp to resell Slowaris x86 has been obvious from the Carly days, so you can rest assured that hp negotiated from a position of strength. I'm wondering if Hurd made Schwartz crawl around on the floor and bark like a dog before signing.
On a final note, TPM also needs to read up on his blades history, seeing as the hp blades range have always been just that, hp products, and not Compaq orphans. Yes, Compaq was the dominant racked x86 server vendor, but it is now hp that are the dominant blades vendor. I can't believe there's enough keeping TPM from doing proper research now that The UNIX Guardian has popped its clogs.
"I certainly don't advocate Windows or Linux as the only platforms on which to deploy new applications"
Why not? Beyond really specialist apps, those platforms will run pretty much everything in any kind of availability mode you want. Restricting yourself to just the two OSs keeps the support costs lower whilst still giving you some leverage for negotiation....
"hp-ux has always had scaling and performance advantages over ProLiant"...
So software scales better than hardware - eh? HP-UX isn't even available for Proliant so how can you compare HP-UX with anything? HP won't port their own OS to their own server range which must mean that they know it isn't up to the job compared to Linux, Solaris, Windows etc.
Sun's Project Copy Linux - oh you mean DTrace, ZFS, etc that's right Matt, they were all copies of stuff done by Linux first. Sure there is technology going both ways but let's just state facts not fiction.
Have HP copied ZFS into HP-UX yet - no and they won't either because HP-UX isn't going anywhere. Itanium is a niche and there is no getting away from it.
Matt, your whole "Slowaris" thing is really pathetic. Are you actually capable of formulating an argument without abusing things? You are the El Reg equivalent of the playground "my Dad's bigger than your Dad".
Who needs Solaris and HP-UX anyway when you can have AIX or Linux on the fastest CPU cores available today on the high end and Windows or Linux on x86 and x64 on the low end?
somebody should have read the latest icd report before writing this article: "Sun was the only top 5 server vendor to experience positive x86 server revenue growth in the quarter – growing factory revenue 21.3% – and gaining x86 market share in the process."
and with respect to blades: "Sun, Dell, and Fujitsu/Fujitsu-Siemens all significantly outperformed the market with year-over-year revenue growth of more than 60% respectively."
also last time I looked hp-ux wasn't exactly the cradle of innovation, much less creating any developer excitement.
it's interesting times with tukwila and 2 new sparc chips coming this year. power7 is also around the corner.
"But maybe TPM should also know that hp's Integrity range scales from two-way blades to Superdome, and all segments are selling well even if Integrity is most dominant in the high-end"
Which is pretty amazing considering that none of the Itanium systems selling today are chip upgradable to Tukwilla if it ships/when it slips later on this year. Or are HP quietly not telling their customers about that?
"Because hp-ux has always had scaling and performance advantages over ProLiant"
The Beckton Core i7 chip will change all that. Using the same 4 x QuickPath interconnect technology as Itanium, with more cores, threads and a higher clock speed than Itanium the scales are going to shift completely the other way.
"In fact, hp-ux's public roadmap goes further into the future than Slowaris, Windows or AIX, and the reason is because customer demand is still very strong."
That's because it's going to take that long for hp-ux to catch up with the rest of the industry. For example, how many more years do we have to wait for a proper cluster aware LVM that doesn't involve hacking with mknod and copying map files around, and that we don't have to pay extra for? Ref Linux to see how it _should_ be done. How much longer do we have to wait for Integrity Virtual Machines to get support for Live Migration, when the rest of the industry licked that problem years ago? And when are HP going to start including modern tools and utilities as part of the base OS to make life easier for admins? Would it really kill anyone to include Bash and Enahnced VIM for example? hp-ux feels old and dusty compared to a modern Linux distro. And don't get me started on the shocking price of per-cpu licenses.
Will sun end up selling software and not selling hardware?
TPM writes an article not entirely complimentary to Sun nor HP, yet MB insists on whining about the HP slights and the article not being harsh enough on Sun. Typical MB FUD spreading. No technical information and he just repeats what he reads from HP's FUD pieces. Get an original thought Matt.
I don't see any Sun fanboys screaming about the unfairness of the article, yet Matt is very defensive. Very interesting... and very telling...
RE: @Matt Bryant by AC SUnshiner Number 1
"...HP won't port their own OS to their own server range which must mean that they know it isn't up to the job compared to Linux, Solaris, Windows etc...." So I assume you'll be pointing the same argument at IBM's AIX and z/OS then? Probably not. I'll make it really easy for you to understand - hp-ux has been developed for the datacenter, it is the high-end OS of choice because it does those business critical jobs really well, and whilst Windows and Linux have come on a long way (especailly on hp servers), they still don' match hp-ux. The market figures say it clearly, hp-ux is taking marketshare from Slowaris in the high-end, which is the most lucrative part of the business. Hp already dominate the Wintel/Lintel market with ProLiant, they don't need to port hp-ux to x64. Sun had to port Slowaris to x64 seeing as SPARC has been such a development fiasco.
"....Sun's Project Copy Linux - oh you mean DTrace, ZFS, etc that's right Matt, they were all copies of stuff done by Linux first....." Well, ZFS is a copy of ONTAP, not done by Linux but still a copy. And the much hyped dtrace is just Sun finally catching up in one area with the tools hp-ux has had for years. In the rest of the management game Slowaris is still years behind. And in trying to follow in Red Hat's footsteps Sun is a decade behind, especially in areas like drivers.
"......Have HP copied ZFS into HP-UX yet - no and they won't either because HP-UX isn't going anywhere...." As I have explained to Sunshiners in numerous posts before, hp-ux doesn't need ZFS, they have a better relationship with Veritas than Sun and real storage technologies. As Linus Torvalds said, ONTAP is much more interesting anyway. Whilst SUnshiners like to drivel on about ZFS, it has won Sun nothing other than a courtcase with NetApp.
".... Itanium is a niche and there is no getting away from it...." Lol, do you mean that nice, profit-making niche of the high-end UNIX space? The one Sun is losing share in hand over fist? Nice niche, much more proiftable than Niagara's webserving niche. Mind you, hp-ux has no counter to the Slowaris vapourware niche King that is Rock.
"....Matt, your whole "Slowaris" thing is really pathetic. Are you actually capable of formulating an argument without abusing things?...." Much as I enjoy using the Slowaris label, I have to admit it is not of my own making. You see, it comes out of years of poor Solaris on SPARC performance and appeared in the industry years ago. I'm not sure when I first heard it but it was quite common here in the UK by 1998. Much as you may like to pretend otherwise, I am not the only dissatisfied former Sun customer out here, there's plenty of us.
"......You are the El Reg equivalent of the playground "my Dad's bigger than your Dad"." And you're just another Sunshiner desperately trying to ignore the writing on the wall. Maybe you'll get it when it's on a pink slip with your name on it.
RE: AIX, Linux, and Windows
"Who needs Solaris and HP-UX anyway...." Well, it looks like all those companies running business critical applications still need UNIX seeing as the recent figures show the last quarter was the first in a long time that UNIX sales came in higher than Windows. And the UNIX of choice for those business critical roles is hp-ux on Integrity, as shown by hp's gains in the high-end market. However, Slowaris has slumped in the high-end so your query would seem to be right about Sun's UNIX.
RE: No Use for A Name
"somebody should have read the latest icd report before writing this article: "Sun was the only top 5 server vendor to experience positive x86 server revenue growth in the quarter – growing factory revenue 21.3% – and gaining x86 market share in the process."...." Sun grew its tiny x64 share from miniscule to just really small. Meanwhile, hp and IBM made much more impressive sales figures, still lead in market share by large margins, and made profits, something Sun hasn't done for years. I also notice you forgot to mention that the largest growth area, blades, is still completely owned by hp with over 50% share.
"....also last time I looked hp-ux wasn't exactly the cradle of innovation, much less creating any developer excitement....." You obviously haven't looked for a while (if ever). I can't speak for developers but there seem to be a lot of CEOs happy to buy hp-ux for the lucrative high-end roles, much more than Slowaris.
RE: @Matt Bryant by AC Sunshiner Number 2
"....Which is pretty amazing considering that none of the Itanium systems selling today are chip upgradable to Tukwilla if it ships/when it slips later on this year. Or are HP quietly not telling their customers about that?...." Old news, the current Superdome for example has had the same frame developed over almost ten years, so it's time for a switch. Still, it's more consistent than Sun which can't even tell you about the next generation after SPARC64 VII (that's if Fujitsu decide to make it), T2+ (who knows what that will need, or if Sun will have the money to develop it), or the follow on to Rock (beyond vapourware seeing as Sun can't even tell us about the future for the planned Rock boxes). For years hp has given customers in-box upgrades, but now is the time for a switch to take advantage of the new QPI.
"....That's because it's going to take that long for hp-ux to catch up with the rest of the industry....." A simple lie to expose - if hp-ux was behind the industry it wouldn't be the leading OS in the UNIX high-end, where the customer demands for performance, reliability and features are at their most. By the way, did I say that's exactly the area Sun is losing most? Looks like Slowaris is the one with the catching up to do. In answer to the rest of your vacuous schpiel I suggest you actually look at hp-ux 11i v3 as you'd see (i) the Base OE is free with Integrity (ii) the LVM filesystem has proven far superior to anything from Sun in real world use, especially its tight integration with ServiceGuard, the UNIX clustering tool of choice that appeared four years before Sun brought out their unstable and feature-lacking clustering (iii) hp-ux can be completely managed from a browser, including the creation of filesystems and clusters, which gives the lie to your lie of "having to hack" anything. Ease of management for hp-ux is one of the reasons we have replaced all our business critical SPARC kit with hp-ux on Integrity, we just don't need to spend as much time firefighting and messing around on the CLI as we did with Slowaris. What edge apps we do have left on Slowaris are all planned for migration to Linux, and on proLiant and xSeries, definately not Galaxy.
/still pointing and laughing!
Could not let this one pass. The late nineties is when the 64 Processor E10K came out. During the period before "SunCluster" was released a product called "PDB" is what was used to support Oracle OPS.
"Much as I enjoy using the Slowaris label, I have to admit it is not of my own making. You see, it comes out of years of poor Solaris on SPARC performance and appeared in the industry years ago. I'm not sure when I first heard it but it was quite common here in the UK by 1998."
Actually, in Silly Con Valley, we were calling it slowaris when it ran on Motorola back in 1989 or thereabouts. I know the history doesn't agree with me, but I have a dual-pedestal 68030 based 3/470 "Pegasus" named "slowaris" that has been running SunOS 4.x for nearly twenty years. Even the Sun reps called it slowaris before 1990 ... I think "Solaris" was a not-quite-real working name for late versions of SunOS4.1.x, which only became the real company-wide name with the release of SunOS5.0 ...
Regardless, corporate UNIX-like OSes only make sense when you absolutely have to have proprietary tweaks for specific hardware that you can't get elsewhere. BSD and Linux can handle most needs. To suggest otherwise, and believe it, is pretty much pissing into the wind. But whatever. Follow your bliss.
The 3/470 is my personal, private legacy system ... it's my family's FTP, email and Web server & is buried under Bryant Street in Palo Alto. It turned up on my doorstop 19 years ago, so I used it to give my family a communications alternative to what was then considered "normal". I have a headless Pentium laptop running BSD ready to take over if it ever fails. The laptop is the third fall-over box ... the first two died on me, but the 68030 & keeps chugging along.
& is buried under Bryant Street in Palo Alto
Where on Bryant Street? I used to live on Ramona (Circle). And how did you bury it?
I have to agree with jake, unless you have some special app most things will run just fine on any generic i386/amd64 box running some Linux or BSD. You'd probably go for Linux simply because it can work with almost any weird and wonderful hardware you can throw at it.
On the Slowaris thing.... on 32bit SPARCs Linux is a good deal slower than Solaris or NetBSD. Something about the MMU code or something I seem to remember. There again 32bit sparcs were slow as shit in the first place.
Contrary to Matt's comments about Itanium's importance, I view this as both Sun and HP confirming the increasing dominance of commodity hardware. The author is correct in highlighting Sun's about-face on resurrecting Solaris on x64, where it can continue to live on - independent of the fate of SPARC as a CPU architecture.
As to HP Integrity's recent up quarter, that likely has much to do with PA-RISC customers refreshing 5 year old soon-to-be-obsolete PA boxen with the latest generation of HP-UX Itanium gear, under pressure from HP (due to their likely use of escalating maintenance fees to encourage migration over to Itanic-TNG). Will be interesting to see what happens over the entire 2009 period.
I find large ISV software benchmarks to be useful when cutting through the fog in these types of discussions - SAP's 2-tier SD Benchmark being one of the best for me (found here: http://www.sap.com/solutions/benchmark/sd.epx ). SAP boils down benchmark performance to "SAPS", handy for making comparisons. One of the more interesting comparisons is as follows ( ["benchmark number"] ):  HP rx6600-8Core / HPUX=10,780 SAPS / versus /  HP rx6600-8Core / Win2K3 Svr=8,680 SAPS / versus /  HP DL585-8Core-AMD / Win2k3 Svr=10,500 SAPS. This shows two things - (1) that the DL585 (AMD x64) almost equals the rx6600 (Itanium) in terms of benchmark performance while likely costing significantly less, and (2) that the SAME rx6600 SERVER loses 19.5% of its performance value when running the non-HP-UX workload. So much for the "synergy" of being able to run both HP-UX and Windows together on Itanium. Why pay more to consolidate workloads onto Itanium, if it also means accepting lower price/performance?
The second interesting comparison would be between these three benchmarks:  Sun T5240-16Core / Solaris=20,090 SAPS / versus /  SunBlade x8440-16Core-AMD / Solaris=17,850 SAPS / versus /  HP BL685c-16Core-AMD / Win2k3 Svr=17,650 SAPS. First, the 16 core UltraSPARC T2 Plus CPUs drive slightly better results (12.5%) than the AMD CPUs when running Solaris. Second, the SunBlade running Solaris is roughly equal to a similar HP Blade server running Win2k3 Server. Thus as long as Sun can crank out SPARC CPUs that beat the commodity CPUs, they can justifiy keeping SPARC around; at the same time, having Solaris running on as many x64 vendor products as possible gives them a "Plan B" HP doesn't have at this point for HP-UX.
In short, Sun now has two different OS strategies going: (1) running Solaris-for-x64 , Linux, and Windows Server on their commodity-CPU servers, and (2) enabling Solaris-for-x64 to run on all the key commodity vendor platforms from their competitors, along with Solaris-for-SPARC on their own and Fujitsu's SPARC-based machines. You can fault their logic, but arguing that Itanium should be a part of that picture makes as much sense as IBM resurrecting its AIX port to Itanic. Itanium is effectively now the same kind of "niche" product as IBM's Power architecture; and as Intel and AMD continue the current "core wars", the niches will get narrower.
Which brings me to my last point. As we move up to Power and Itanium CPUs with 4, 6, or 8 Cores competing with AMD and XEON commodity CPUs having similar configurations, it will be all about how the OS effectively leverages that power, not the instruction set (EPIC's VLIW approach now appears to be quite "quaint" from my perspective; a product of a bygone era). So for the Solaris haters out there, I would like to remind you how much experience SUN has built by scaling their OS over lots and lots of CPU cores/threads over the past 10 or so years. Their porting of Solaris to x64 may not save SPARC, but it just might save Solaris, which may be the only OS on commodity CPU servers with extensive commercial experience with large-scale SMP (32 to 64 Cores) for the next couple of years. Windows Server 2008 will have this capability and experience eventually, but not for a while; same with Linux (not talking "theoretical" scalability here; how many Windows and Linux severs in actual use today exceed 4 CPU sockets, thus limiting them to 16 CPU cores?). Meanwhile, I doubt that either HP-UX or AIX take this route; this issue is a major differentiator for both vendors until Windows and Linux catch up.
"Alien" because I don't work for either HP or SUN.
If Mr T. read the reg I think he would post "Quit your jibber jabber fool !"
As you clearly stated your views are based in the past and you have no intentions for forming clear unbiased views, your current views are based on your poorly understood out of date marketing rubbish.
Your posts have become boring as its the same old tripe for anything to do with Sun, I use different systems myself and your posts are turning me more anti HP. The loan rantings of a desperate person telling everyone they are wrong and he is right is sad.
Good luck Matt My'rant....
I'm predicting hordes of frothing Sunshiner mobs with, torches and pitchforks, converging on Bryant Street right about now...... I often wonder what the origin of the Slowaris tag is, someone told me it came out of IBM marketing back in the day. I'd be interested if anyone knows for sure.
Agreed, a lot of what we used to run on UNIX of one flavour or another has moved to Linux, some to Windows, but there are still the core bizz crit bits the board will only run on tried-and-tested UNIX stacks like hp-ux on Integrity. Given time and resources, I'd say we could replace 99% of everything with Linux, but I just don't see it happening. Looking back, the big surprise is how much Windows has actually penetrated our datacenters, particularly with MS SQL replacing a lot of departmental Oracle on UNIX. Ten or so years ago I would have predicted the Linux invasion but not so much of the Windows, we had a rule of "no Windows in the datacenter" then. BSD, especially FreeBSD, did have a look in but was lost in the enthusiasm for Linux, especially Red Hat. And despite the best efforts of our semi-tamed Penguinistas, it looks like Windows is pretty secure on the desktop, despite the regular virus scares and expense.
RE: RE: Late Nineties and SunCluster
Ah, Sun PDB for OPS, I remember it well. Especially the crashes, the network panics, and the finger-pointing exercises between Oracle, Sun and EMC. Yes, Early ServiceGuard wasn't perfect and had its bugs, but it was like an oasis of calm compared to Sun's dismal efforts. Sorry, but the idea of comparing PBD or even SPARC Cluster 2 to even the early versions of ServiceGuard is likely to induce fits of manical laughter amongst those of us that suffered that Sun product. I still remember how the hp salesgrunts used to salivate whenever we said a new project had a clsutering requirement, as they knew ServiceGuard would put them in pole position competing against Slowaris.
As HP continues to engage SUN as a partner with increased visibility, Matt still can't spell Solaris.
It seems Matt can't read, either. The article was about HP, yet his comment was mostly about SUN.
Perhaps Matt Bryant's problem with spelling and reading are related?
"Where on Bryant Street? I used to live on Ramona (Circle)."
Just south of University, in the old telco CO that became a major peering point back in the old days. Anyone other than me remember BARRNET?
"And how did you bury it?"
I have friends in low places, and sometimes write odd things into contracts. Before the Sun, I had an AT&T 3B1 "Unix PC" in the same location for a few years, prior to that it was an LSI11 that had been running since before Flag Day. That connection *MIGHT* be the very first that we would now call "coloed".
If anyone can place a name and face on me from this info, kindly keep it to yourself. Ta. On the other hand, if you can place a name on me, you can find me in Sonoma. Feel free :-)
Or are they the same Sunshiner? Anyway, on with the fun!
RE: @ Matt Bryant - Quit your jibber jabber fool !
"If Mr T. read the reg I think he would post "Quit your jibber jabber fool !"...." I'm not surprised to find that your basis for a technical argument is a fictional aggressive character with limited intersocial skills. Oh, wait a sec - what technical argument? At this point I've given up even trying to look surprised by you Sunshiners.
"....As you clearly stated your views are based in the past and you have no intentions for forming clear unbiased views, your current views are based on your poorly understood out of date marketing rubbish...." As you clearly failed to read, we don't just take any company's word, we make them go to shootouts with real data in our environment. Which means I regularly get to see the best Sun can provide against the best from hp and IBM. In the last five years Sun has not one a single shootout, not even close. Which means I probably have a better view than the majority of the frothing Sunshiners that post here. In those five years of comparative benching, despite our request that the companies just provide info on their own products and despite their expansive promises, it is always the Sun representatives that are guilty of continually FUDing. I have posted this in my previous posts, but I assume you were too busy ranting to read it.
"....Your posts have become boring as its the same old tripe for anything to do with Sun...." I'll tell you what - you Sunshiners stop posting FUD and I won't feel the need to expose you for the liars you are. You can live on in your little bubble of denial without tainting the rest of the IT world with your FUD, and the rest of us can get on with watching the slow-motion carwreck that is Sun. Winners all round!
"....I use different systems myself and your posts are turning me more anti HP....." Strange how you feel so driven to defend Sun then..... I assume you use "different systems" because Sun couldn't meet your requirements either?
"....The loan rantings of a desperate person telling everyone they are wrong and he is right is sad....." Hey, they're not on loan, I give 'em away for free. If you feel the need to pay for them - Sunshiners seem to think it's right that you have to pay dearly at some point - then please go to www.linux.org and click on the "more" link in the "Make a donation" section. Whilst you're there, I suggest you also have a look at http://www.linux.org/lessons/beginner/index.html to prepare you for when that pink slip arrives.
/Sides starting to ache from all the laughing!
OK, disclaimer time for the incredibly obtuse Sunshiners - try and read and comprehend this time, dweebs.
I do not work for hp, IBM, Dell or any other vendor.
I am not paid by hp, IBM, Dell or any other vendor.
I do not work for a reseller.
I do not work for a marketing company.
I am not a bitter ex-Sun, ex-FSC or related company employee.
I have never applied for a job with and have never been rejected for a job with Sun, FSC or any related company.
You all need to learn Linux - fast!
Solaris already runs on X86_64 hardware, it is easy to port drivers for the proliant line.
If you have seen the new Suns Openstorage 7XX0 boxes. ZFS, Fishworks +++, The most sexiest NAS ever. Getting HPUX up to this level is completely impossible without a Lunar landing budget.
It is beyond me why anyone bring the Compaq name into this. HP kept the proliant name on their mid range X86 servers, which are well fitted for the solaris OS.
My Coat is the one that says 'Intel Please put Itanic to rest!!'
(I tried posting this, or something similar yesterday, but it seems to have gotten lost in the aether)
"I'm predicting hordes of frothing Sunshiner mobs with, torches and pitchforks, converging on Bryant Street right about now..."
Whatever. As a founding father of the vi/EMACS wars, and later pretty involved in the C/C++ "debates" on Usenet, I don't have time to be religious about technology anymore. It's just a big waste of time ... Basically, this old fart's opinion is that whatever kit is right for the job is the correct kit to apply to that job, regardless of brands & logos.
"... I often wonder what the origin of the Slowaris tag is, someone told me it came out of IBM marketing back in the day. I'd be interested if anyone knows for sure."
Back in the eighties, I worked for a company that supplied hardware for the backbone of IBM's internal voice & data network. We shipped Sun workstations along with our kit, for network monitoring and statistical analysis of network traffic. (Yes, IBM used Sun computers internally, it was the only platform our network software ran on at that time.) We were a cutting edge company, and we worked with other cutting edge companies to ensure "in the field" compatibility. We were a pilot-build test site for the likes of cisco & Sun. As a result, we often had IBM and Sun and cisco engineers in the building.
In the mid-late 1980s, Sun rolled a rev on SunOS and asked me to be a Pilot site for their new product ... The OS went from 4.0.x to 4.1.x ... along with the rev came a new handle; it was nicknamed Solaris. It also brought usable Sparc boxen into the fold (the SPARCstation 1 running SunOS4.0.3c was a useless POS). I do not remember if any of the documentation was labeled Solaris, but that was what the Sun guys (and gals ... Hi, NancyN!) were calling 4.1.x.
The early Sparcs running 4.1.x were absolute pigs compared to 4.0.x or 4.1.x on Motorola, and within a couple weeks, maybe even a couple days, everyone was calling the "new" OS+Sparc "Slowaris". Even the Sun reps. Did it start at my company? I don't know. But it wasn't marketing ... it was engineers that started it. I don't think IBM marketing even knew it existed at that point. The 2/, 3/ and 4/ machines were always called "sun stations", and never negatively.
Come to think of it, after all these years, Sun was probably working on Sun5.0 "Solaris" at the time ... I'll bet somebody in Sun marketing got wind of the new 5.0 OS, and jumped to the conclusion that when 4.1.0 was released it was the new Solaris release ... But we'll probably never know for sure.
The dates above are from memory, and I might be concatenating time again ... but the events are as described. Again, if anyone recognizes me, please keep it to yourself ... or feel free to get in touch. You know how to find me.
(Again, I tried posting this, or something similar yesterday. I wonder what I'm doing wrong.)
"You all need to learn Linux - fast!"
Not necessarily Linux, but rather FOSS. The kernel is immaterial in this context.
Distributions will come and go, and it doesn't really matter if they are based on Linux, BSD, Plan9, Minix, ai/x, hp/ux, solaris, osx or whatever. Or even WinDOS[tm][c][r], for that matter. All are going to eventually go away (except Slackware, which will live on forever of course <sfsf>). But by its very nature FOSS will be around indefinitely. Folks who ignore this fact are already falling behind the technological curve. Like it or hate it, FOSS has real, mature solutions for corporate computing. And it continues to get better, unlike some Corporate OSes I could name.
Is FOSS usable by John and Jane Q. Public? The answer is a highly qualified "maybe".
For example, I was constantly summoned to my Aunt's house (93-year-young family matriarch) to "Fix my Windows, please" ... I got tired of it, and tried to convince her to try the cut-down version of Slackware that I built for my mom. She wouldn't have anything to do with it. So I cleared some space in her home office, and setup the Slack box, just to see what would happen.
After about a month of no support calls from her, I dropped by to see what was up. The Windows box had gotten fouled up, and she was expecting some email from her sister in Finland ... so out of desperation, she tried the Slack box. Seeing as she only uses a word processor, a picture viewer and a web browser, those are the only icons on her desktop. She already used Mozilla on Windows, a word processor is a word processor (for most people) and the same for a picture viewer, so the change was fairly easy for her.
That was over a year ago now. A couple months ago she asked me to get rid of "that dratted old thing" ... I couldn't convince her that I could install the same variation of Slack on her old box, with it's faster CPU, more RAM, bigger HD etc. ... She is absolutely convinced that the hardware & software are somehow married & the old box was a lemon ... but she's completely in love with "this new version of Windows that my nephew gave me" ...
In a nutshell, Linux CAN be used by folks with little or no knowledge of how it works ... but only if the box is setup by someone who understands their needs AND their lack of technical ability and is willing to put in the effort to create a FOSS solution for that user.
None-of-the-Above said, "(1) that the DL585 (AMD x64) almost equals the rx6600 (Itanium) in terms of benchmark performance while likely costing significantly less, and (2) that the SAME rx6600 SERVER loses 19.5% of its performance value when running the non-HP-UX workload."
If this is the case with AMD, the question in my mind is how much more will Itanium suffer in performance in comparison to the new 6 core Intel CPU's?
The price-performance of AMD to Itanium places HP-UX at a severe disadvantage - I am not sure what value HP-UX under Itanium brings to consumers.
None-of-the-Above said, "the 16 core UltraSPARC T2 Plus CPUs drive slightly better results (12.5%) than the AMD CPUs when running Solaris. Second, the SunBlade running Solaris is roughly equal to a similar HP Blade server running Win2k3 Server. Thus as long as Sun can crank out SPARC CPUs that beat the commodity CPUs, they can justifiy keeping SPARC around"
If applications were tested with encryption enabled, the T2 processors would pull away with a much higher performance margin since the other processors would start to plummet while the T2 would hold its superior performance metrics.
I think the market is hoping to see a lower price T to be released, in conjunction with a higher performance T version. More diversity in the T line would be nice, to help SPARC continue to compete with x64.
Price/Performance are the same challenge Itanium has been experiencing with x64, as SPARC is dealing with x64. CPU agnostic Solaris could make a nice bridge for HP, as it has done for SUN.
<Sigh> Are you Sunshiners STILL trying with the cherry-picked SAP benchmark figures, even after you've been shown up? OK, if you want to be hoisted on your own petard, away we go!
The DL585 bench session was done in November 2007 whereas the rx6600 bench was done in 2006. The rx6600 was using the old 9050 Itaniums, which had been replaced with faster models by the time the DL bench was done. So outdated Itanium still beat the then latest Opteron.
Now for the hoisting! Seeing as you seem so eager to throw Opteron SAP benchmark figures around, shall we look at benchmark session 2008062? Yes, the benchmark session that haunts Sunshiners everywhere! In November 2008 the poor ickle M3000 with the latest SPARCVII chip stumbled to a derisory 4180 SAPS. Care to look at bench session 2005024? Yes, that's the old hp BL25p Opteron blade from 2005 hitting 4880 SAPS. Mind you, it could have been worse, seeing as the BL25p also bested the later Niagara-based T2000 (4780 SAPS, December 2005, bench 2005047) and the Sunfire X4200 (4180 SAPS, December 2006, bench 2006087). Indeed, the BL25p G2 model totally creamed them when it posted 5250 SAPS (August 2006, bench 2006064). So it looks like SPARC is completly irrellevant whan compared to Opteron, they just can't even compete with the older versions!
But for real fun let's look at some more recent Opteron-based hp bench sessions. How about last December's 2008064, where the hp ProLiant DL785 G5 8-socket posted 35400 SAPS, compared to the October 2008 bench session 2008061, where the similar Sun Fire X4600M2 8-socket only managed 29670 SAPS. Yes, I know the Sunshiners will start whining about how the hp server had the 2.7GHz 8384 4-core Opterons, whereas Sun's server had to make do with the 2.5GHz 8360 4-core Opterons, but it's strange that 8% jump in CPU clock gave a 20% jump in performance. Oh, hold on a sec, I see the differentiator - the Sun box was struggling with Slowaris 10, whereas the DL was humming along with SuSE Linux. Looks like SPARC isn't just shown up by the SAP benchmarks, Slowaris on x64 is also exposed as the complete shadow of a real Linux.
And to close, may fave de-Sunshiner! Bench 2006089, from December 2006, where a 64-core Superdome using those old 1.6GHz 9050 dual-core Itaniums still posted 152530 SAPS, compared to the latest Sun M9000 with 64 Fujitsu dual-core 2.4GHz SPARCVIIs, which only managed 129420 SAPS in November 2008! Even with the latest M9000 being stacked with double the RAM of the old model Superdome, the latest Sun kit couldn't compete. Core for core, the latest SPARCVII got thrashed by a three-year-old Itanium. No wonder the Sunshiners are FUDing Tukzilla so hard.
David's self-hoisting is really beyond even the drivelling Sunshiner norm, so I've decided to honour him as the first with a term I've coined for those, so dazzled by the Sunshine coming out of the Sun Rock camp, that their ability to objectively consider anything is about the level of the comatose. Given Sun's codenaming the new Rock vapourware as "Nova", please all point and laugh at the David "Novatose" Halko, no doubt the first of many!
"OK, if you want to be hoisted on your own petard"
That's pronounced "hoist by your own petard". Look it up. Pet peeve.
 or "hoise", if you want to be archaic & petulant.
Sorry, Ms Bee ... I'll ignore this thread now ;-)
Correction gratefully noted. Not to worry, I'm sure Novatose is still trying to work out what a pertard is. Please don't feel the need to leave, some of us value the contribution of self-proclaimed "old farts". Can I ask your opinion as to whether you think Slowaris x86 has a chance against established Linux distributions such as Red Hat?
"Please don't feel the need to leave,"
I don't feel a need to leave. I feel a need to avoid making work for Ms Bee & anyone else involved in moderating this place, in a thread that is basically two or four people nattering. Take it to Usenet; that's what Usenet's for. This will be my last post in this thread, unless I see a need to make a point similar to this paragraph.
However, now I've made that point ...
"some of us value the contribution of self-proclaimed "old farts"."
::heh:: I try. But I'm not infallible ... Get a second and/or fifth source for anything I type(o).
"Can I ask your opinion as to whether you think Slowaris x86 has a chance against established Linux distributions such as Red Hat?"
Probably not the answer you are looking for, but I've installed more Solaris x86 in the last couple years than I have RedHat. In fact, outside of testing in my lab, I haven't done a single RedHat install (ever, in any environment) ... and I have installed Solaris x86 in place of existing RedHat many times in corporate environments.
THAT said, I invite you to re-read what I wrote above (4th March 2009 20:25).
Matt ~ "<Sigh> Are you Sunshiners STILL trying with the cherry-picked SAP benchmark figures, even after you've been shown up?"
David Halko did not post any SAP benchmarks, it was None-of-the-Above! - - - Better learn how to read, Matt.
Matt ~ "compared to the latest Sun M9000 with 64 Fujitsu dual-core 2.4GHz SPARCVIIs"
The latest SunM9000 SPARC VII uses a quad-core processor. - - - Better learn a little bit about architecture, Matt.
Matt ~ "hp's Integrity... is most dominant in the high-end. Sun's big problem is it is losing share in that high-margin high-end"
None-of-the-Above shows sun dominating the high-end benchmarks with SAP.
Matt ~ "...M3000 with the latest SPARCVII chip.. Niagara-based T2000..."
Did you change the topic to low end boxes because HP was shown inadequate next to sun at the high-end? - - - Better learn to stay on task, Matt.
"....David Halko did not post any SAP benchmarks, it was None-of-the-Above! - - - Better learn how to read, Matt...." No, he just repeated the same vacuous post with the same blithering ineptitude as when he posted EXACTLY the same figures in a previous post - gee, I wonder if he's also posting as AC and None-of-the-Above to make the Sunshiner response look bigger?
"....The latest SunM9000 SPARC VII uses a quad-core processor. - - - Better learn a little bit about architecture, Matt...." Apologies for my typo, but do you deny that core for core, even though the SPARC64 VII cores had a 50% faster clock and double the memory of the Itaniums tested, the three year old Itaniums still beat the latest SPARC64s with ease? Care to suggest how four-core SPARC64, even with the mild speedbump of VII+, will stand even the slightest chance against a four-core Itanium like Tukzilla, where each core will be much more powerful than the three year old 9050 cores? Quick and simple answer - they won't. The SPARC high-end is dying fast because the Sun customer base has realised they can get better performance with better reliability from hp.
"...None-of-the-Above shows sun dominating the high-end benchmarks with SAP....." No, but SAP on UNIX is mainly an enterprise-level product, because lower down it is already running on Windows or Linux, neither of which SPARC can do as a viable commercial offering. If the Sun kit is obviously lagging on SAP, an enterprise product, then it suggests SPARC Slowaris is also lagging on other enterprise stacks as well, an idea which would seem to be supported by the loss in the high-end very evident in Sun's marketshare figures. You Sunshiners introduced the SAP benches, argue amongst yourselves why, I'm just pointing out how stupid you guys were to introduce them in the first place.
"....Did you change the topic to low end boxes because HP was shown inadequate next to sun at the high-end? - - - Better learn to stay on task, Matt....." No, I brought in the low-end comparisons to show how SPARC is lagging across the whole range, especially after Novatose said Itanium was somehow irrellevant seeing as the old 9050 chips could only just beat newer Opterons. What better way to show SPARC's irrellevance than to show how even old Opterons beat the latest four-core SPARC gear? The DL785 vs X4600M2 comparison was just to highlight that not only do hp do the whole Opteron thing better than Sun, but that Linux also performs better on Opteron than Slowaris x84. I'd say that's pretty much task complete, but thanks for the opportunity to point out once again that Sun servers in all forms are the biggest performance laggards in the market.
Slackware? Oh come on! The FreeBSD trolls are laughing up their sleaves at that one! :P
The most important parts of Matt's CV, from a previous post:
- I do not work for hp, IBM, Dell or any other vendor
- I am not paid by hp, IBM, Dell or any other vendor.
- I do not work for a reseller.
- I do not work for a marketing company.
- I am not a bitter ex-Sun, ex-FSC or related company employee.
- I have never applied for a job
In other words, no experience in what he talks about.
I sense the usual complete-lack-of-a-technical-argument-but-instead-a-lame-insult post of the ardent Sunshiner. Tell me, if my having worked with enterprise UNIX, Linux and Windows for over a decade (and that's a decade of real enterprise projects, before that I also worked for smaller companies and even a college), in your eyes means "In other words, no experience in what he talks about", I assume that is also the attitude you hold to the rest of your customers then? Is it a case of you really thinking only Sun employees or their resellers have the right skills and knowledge, and your customers are just idiots that should act lucky to buy your beloved Sun products? Well, all I can say is I really hope you do work for Sun, for a Sun reseller, or for a marketing company working for Sun, as you seriously sound like you shouldn't be in the industry at all.
Matt Bryant ~ "I sense the usual complete-lack-of-a-technical-argument-but-instead-a-lame-insult post..."
It looks like it was a cut and paste from your own post on "Tuesday 3rd March 2009 09:31 GMT"
If I could figure it out, not sure why you could not.
Now, do you remember what I posted about the Sun sales tactic, the one where they would avoid a technical showdown by trying to belittle anyone that didn't toe their line....? Looks like the resident Sunshiners are suffering from that same old Sunshiner lack of new ideas that has driven Sun down to a shadow of the company it once was.
PS: Jake - Slackware is great if you have an accomplished admin to configure it for you, whereas even most grannies can cope with Windows out of the box. My preference for complete PC noobs and the blue-rinse set would be Xandros or Linspire. If they really have the WIndows bug already then maybe a Slackware-based XP clone like Darkstar Linux might be an idea.