SPARC is dead, ROCK is history, SPARC64 is Fujitsu tech
Expect that $1.5B to $2B to come from extortion of Sun's customer base.
Sun Microsystems' soon-to-be ex-chief executive has been painting his company's acquisition by database giant Oracle in positive tones. Jonathan Schwartz called the proposed $5.6bn deal a "fantastic day for Sun's customers, developers, partners and employees across the globe". It will also "redefine" important boundaries in IT …
Don't underestimate the influence that Sun will have on Oracle. Although I mostly agree that Oracle will dominate thought leadership here :-|
I'm just glad it's not IBM taking over. That would have been an ugly death to witness.
I guess Larry has always wanted to play with Scott's toys. Now he gets to keep them!
"Open source will continue at Oracle - along with Java. It could even profit. Just don't expect it to help anybody else, though."
And why not? If Oracle continues open source (as you say), it will inevitably benefit others as well. It is in the very nature of the phenomenon.
The only way Oracle can seriously harm the old Sun projects is by turning them into proprietary closed source. But it can do this only to those projects where it owns all copyrights, and even to them only in some new version that has a changed license. The version prior to this remains free, and can be turned into a community fork, or taken up by another vendor like Red Hat or Novell (the latter already effectively has its own fork of OpenOffice.org). When a piece of open source is popular enough, it cannot be killed by an acquisition.
You have completely missed the point. SunNIOracle is a Virtualisation Company for Networks InterNetworking Java Applications and ESPecial Algorithm ProgramMING Projects in the Open Source Internet Service ProVision Environment of Crack Compiler CodeXXXX .... Intellectual Property Supply for Binary Conversion to Pixellated Content Transmission and BroadBandCasting.
Sublimely EduTaining IntelAIgents Presenting with Pictures the Future Directions Virtually Available for Present Realities. Choice you can Believe in because IT and Media can Build IT with SMART Networking InfraStructures ..... Civil CyberSpace Control Nodes Sharing the Universal Master Plan ....... with Big Blue Prints ..... and XSSXXXXual Scripting for Engaging and Energising Vitality.
And if that is not what SunNIOracle are Tuned into/are all about, then IT must be what NIRobotIQs are Turned onto/Immersed in, and although such would make for an Impressive Troika/Trilateral Proposition, which would not so much Create a Monopolising Presence as Introduce a NeuReal Concept and Realise IT via Global Operating Devices and Sharing Information Systems for Advancing IntelAIgents ... [In MODified Phorms, Mutual Intelligence, MuI7] ..... it may be Best and Better Beta Suited for All to Remain and Retain a Cavalier Independent Disposition in the Fields of Play which would be Constantly Searching for Near Perfect Matches, Made in Heaven, for such is the Field of Choice.
Oracle knows how to run and manage software business. Very soon the Sunacle JVM would be THE JVM (unless some corporate action blocks the evolution).
I am not so sure about the hardware business. Sure Solaris will stay.. but on what platform ?
SPARC may see a pre-mortem burial but - as everyone seems to note - that's better than the entire Sun getting buried.
Interesting to see how Oracle to manage it's partners ? [other hardware vendors]. The business on Sun platforms is less than half for Oracle's total revenue. Oracle can't afford to antagonize it's old partners (including 3rd leg in the affair: IBM).
Sun's marketshare ? It will dwindle.. there is some uncertainity even now. And some CIOs already made up their mind about moving ON when the saga was going on. Things would have worked out better if IBM was never in the picture. Going forward Sunacle needs to up it's publicity machinery.. announce measures like roadmaps of sun's products to restore customer faith.
Getting rid of Sun middle management - however would be the hardest thing for Oracle. Divide and rule should be the name of the game. Break Sun into smaller unit which should re-attach to different Oracle Units. Eg: Solaris should move under the database group. Mysql, OTOH, should stay away from database else some sql gpl'd code may end up contaminating the whole oracle product lines.
SPARC product line should move into some division where they slowly kill old products. Sell all the IP (and some teams) to fujitsu. In future, let fujitsu design and manufacture the SPARC/ROCK line. License some processor technology to Intel and get rebate on x86 in future.
What else ? Keep the Sun brandname alive - mainly to push servers.
Paris, coz if she swallows something it's for good.
Anyone thinking what will happen to btrfs now that Oracle will have ZFS. The reasoning for BTRFS was in part due to licensing issues with porting ZFS to linux...
But now Oracle will have both BTRFS and could change licensing of ZFS....
Personally, I think Oracle has probably not thought that far down.
PS... And re those "Civil CyberSpace Control Nodes " .... they aint a million miles away from Cloud Controls in ESX4, and will always drop back to drag them forward into Rich Universe Applications of the RightScale, for they are Facilitating Hubs for Virtual Operating System Systems. But you were told of that earlier, more than just once, but you must have missed it or ignored it ..... http://amanfrommars.baywords.com/2009/04/20/080420/.
Hello, Brave NeuReal IT World. Wanna Play a Real Game with Global Operating Devices in Control of Power Control Levers with Simple Human Readable Texts into Virtual AIMachinery Protocols for Immaculately Resourced Environments...........Digitally Mastered Binary Landscapes, Media Presented for Swarming with Humanised Assets, Manufactured Products and Population Centres......
A Game brought to you by Special Advanced IntelAIgents Services of which One and All may have Full Knowledge but of which only a Very Few are Ever Fully Aware, such is the Nature of the Reality of Virtual ProgramMING for ITs Projects.
And now y'all know .... and can hardly say you missed it, although ignorance, and I also suppose arrogance, will ably aid all who would choose to ignore it.
Possibly - your eloquent, detailed and obviously highly knowledgable proposition is hard to refute - but consider what a good fit the Ultra T2's (and even the T1's) are for database applications.
For those sorts of loads you really don't need a narrow, fat space-heater of a chip that can run some game-fanboi's wet-dream at 1,000 fps - you want a wide, skinny processor that can give you good performance per watt over a sack-load of threads... and that they do in spades, believe me.
So maybe they'll go - maybe not - maybe sold to HP or someone else down the line; but as a good hardware partner to your core (software) business it might be foolish to simply ignore them.
Just my tuppence worth
This post has been deleted by its author
I could not agree more with the article, however I think that is missing the fundamental reason that is driving Oracle to buy Sun.
Why nobody can see the pattern? Oracle does noy buy Sun, it's buying their customer base. All those Java, all those MySQL paying customers, all those Solaris shops. They can raise their support charges just below the point of making the business case for moving to the competition not workable under less than five years. Job done. Profit. Let the products slowly decay until the next competior comes up. And then simply just buy it.
Look at what are the latest Oracle big acquisitions. Siebel? Peoplesoft? BEA? Java, Netbeans, Glassfish, MySQL will suffer the same fate. When Oracle buys some business it's not buying their products. It's buying the right to go to their customer's door and show some PowerPoint slides where they say that they will keep everything as it is and provide miraculous and impossible migration paths for wathever products they are developing now. Of course, those developments are not important, because if they fail to match the competition they will just buy the competition.
In the meantime, charge their customers wathever they want or else let them die, paying hefty maintenance fees in the process. Look, a company with something like 4 databases (or 5?) 2 ERP, 2 CRM, 3 application servers, and a 22 page product catalog where most of the products are not even worth enough to be sold alone and only are slipped as part of some bundle with the database only to be forgotten and unused. Are you really going to believe any strategic statements from those guys?
The only ray of hope is for those open source products that can be forked, as the article says, and taken over by someone else that is not interested just in share price but in creating innovative solutions that are interesting and helpful to customers.
And please, ignore hardware. Oracle does not care about hardware and never has. SPARC is dead, for all intent and purposes.
"Why nobody can see the pattern? Oracle does noy buy Sun, it's buying their customer base. All those Java, all those MySQL paying customers, all those Solaris shops."
I think, with all due respect to your ego, that it is a no-brainer that this would be a major part of the attraction for Oracle - in fact they as much said so with their previous offers, and this is widely discussed around the world (so not many points for that).
"And please, ignore hardware. "
"Oracle does not care about hardware and never has. "
Mostly true - and I don't think anybody thinks otherwise.
"SPARC is dead, for all intent and purposes."
Nah - well, the pre-Niagara chips are being put out to pasture anyway and have not really been in Sun eye-line for some time now - but the the UltraTx line makes sense if you're shipping a complete enterprise database package as you can offer significant savings on infrastructure, running costs and have excellent transaction throughput . The question of whether you want the overhead of development of them is another matter - but having some leverage there would be a potentially useful thing.
 i'm not a Sun employee btw
Why would Oracle continue to develop BTRFS and essentially give it to every other competitor, when they now have the real thing, ZFS, which gives them a nice differentiating factor compared to Linux ? A good point that was stated in I don't remember which article somewhere this morning.
You might expect Oracle to stop work on BTRFS, and develop solaris for all it's worth (and it's a very very nice Posix platform) as their own, likely with some form of community involvement.
I'm missing Sun already...
Despite all the FUD that Matt guy rattles off, ROCK is too far along to be killed. If transactional memory works as expected, I can't see Oracle wanting to drop it. There are a lot of shops who depend on large Unix servers to run their Oracle DB. Hopefully the Reg will find and leak a few more tests from some of the shops beta testing the new ROCK servers.
For the Niagara Sparc line... This server line does run Oracle & Weblogic very well. The catch is, that Oracle has made it cost prohibitive to do this. Now that they own the hardware, anyone see them changing their licensing for Sparc? Sparc may ramp up! LDOMS also have some cool features to use Oracle RAC. They may now expand on those features.
I can also see Oracle having less interest in BTRFS & Systemtap. They now/will own better products in Solaris, which they can license. I just hope they don't kill off OpenSolaris & some of the other Sun products. I imagine OpenSolaris will stay as it's become Solaris n-1, but it will be some time before we see Oracle's goal after the merger if it happens.
Say it with me, and remember: NetBeans was made open source, and is governed by the "NetBeans Community" -- Sun is the project SPONSOR, but NOT the project OWNER. Oracle can't do diddly squat about NetBeans (and probably wouldn't want to anyway).
If Oracle doesn't want to fund NetBeans, that's fine. Someone else will. Even if nobody does, there will still be forks, and other companies will pick up the slack and offer their own branded versions. Hell, even if that DOESN'T happen, as long as we have the source code we can recompile it for any platform that exists in the future. NetBeans is relatively agnostic about Java versions, as long as you have a working JDK you can direct NetBeans at.
What are you all so worried about? Oracle might as well try to kill the wind. This goes equally for any other open source project Sun had its fingers in. That's the whole POINT of open source.
Oh, and by the way, this goes out to any Eclipse fanboys who happen to be wandering by: Oracle has it's own IDE, JDeveloper, which is better than Eclipse and almost as good as NetBeans. Oracle is NOT going to support Eclipse over JDeveloper. Maybe you should kiss Eclipse goodbye!
Almost everything good related to Java is OPEN SOURCE, and has already been GPL'ed. If Oracle wanted to make it proprietary, they arrived late to the party and the hostess has already gotten drunk and shacked up with hackers.
You can't unring a bell.
This was a very good article!
A few comments...
Gavin Clark writes, "The only thing that's going to get redefined is everything Schwartz and Sun have been working towards for five-plus years: platform independence, open systems, and affordability. Grand ideas Sun was in no position to implement thanks to its inertia and bureaucracy"
This seems like an odd statement. Oracle's applications are nearly all cross-platform. Oracle's applications run on Open Systems. Oracle even has their own version of Linux (which is nearly an Open System - Linux kernel is pretty much controlled by one person and operating system interfaces are not necessarily driven by Open Standards community, but by the Linux community.) Oracle now had been purchasing proprietary platform from an external vendor in the hardware market, it looks like Oracle has finally entered into the Open Systems' Market completely with SUN's Open System line of servers! With MySql, owned by SUN, it seems Oracle has just entered into the affordable market for databases, as well.
Gavin Clark writes, "this would assume Oracle is interested in building an open, platform-neutral, and affordable third-party partner community. It isn't."
Considering that Oracle had embraced Java very early on, which is a platform-neutral, embraced affordable BerkeleyDB, and Oracle has been building applications on everyone's platforms (regardless of hardware & OS vendors, even vendors who competed with Oracle at the database & middleware level) - I find this statement very difficult to agree with without some kind of supporting evidence.
Gavin Clark writes, "The idea of the appliance is an evolution of Oracle's earlier belief in the software stack - application server, database, middleware, and applications - all coming from the same provider. That Schwartz should now endorse this is not a surprise, but it is a shocking U-turn on his work at Sun."
This seems like another odd conclusion. Jonathyn Schwartz purchased StorageTek to get into the Storage Arena and guided SUN to enter into the Open Storage arena with an embedded operating system for an appliance. Under Jonathyn Schwartz, SUN had really worked hard to produce a complete SYSTEMS stack, from the CPU, Hardware, Firmware, Hypervisor, OS, Middleware, Applications - and it was all one vendor (SUN.) There is no U-Turn. Just because someone wants to play nicely with others (using Open Standards) does not mean they reject the idea of supplying a whole solution!!!
Gavin Clark writes, "Despite all that bravado of Oracle becoming a hardware company with last year's Exadata and database machine, the devices are actually built by HP."
It seems like this is a very good fit for SUN - to provide the hardware to Oracle so all sales revenue can flow internally to the same bottom line. This seems like a very good fit.
Sorry, chaps. Oracle isn't going to keep the SPARC alive just because it threads well. Large swathes of their technology strategy - from their Cloud Computing model to the hilariously named Unbreakable Linux - scream "We Love Intel". Heck, they even said exactly that in late 2008, mostly to annoy Sun and point up how market-trailing the UltraSPARC-based server range had become and how overdue Rock was.
Any predications based on technology are fundamentally misunderstanding Oracle's business model and reflect the mindset of Sun inside and out. Sun were always great at engineering but lousy at productization (examples: the game-changing E10K was a Cray product design, and the current M-series are Fujitsu boxes). The expectation that great engineering leads naturally to commercial success is/was a pervasive fallacy. There's a hint that someone got a clue with the Amber Road product direction, but that was very late in the game to start competing with NetApp.
Best path for the SPARC's survival now is a sale to Fuj. Expect Oracle to focus on Sol x86. Because Nehalem blades perform well enough to deliver the vast majority of enterprise workloads, and they readily repurpose for Linux and Windows workloads. So the SPARC's legacy of exceptional multiprocess/multithread performance just ain't that interesting anymore.
Sun always cared about great product, coming from an academic world... Oracle is everything but great products, only money counts... Everything Oracle does is about customer lock in on own products.
IBM would have been a better fit...
So long SUN, thanks for the ride...
For every opinion there is an equal and opposing opinion. NetBeans is "dead" according to one person, but for the guy blogging about the death of open source Java (clueless, given that Java has been made available under the GPL), NetBeans kills Oracle's own products.
The first person to listen to on all of this stuff is Monty, formerly of MySQL AB. Not only was he a Sun insider who knows Sun's product lines, and not only has he had to deal with Oracle before, but he also has a few choice words about the future of MySQL. And no, it isn't "MySQL is dead", either.
Joshua Goodall comments, "Sorry, chaps. Oracle isn't going to keep the SPARC alive just because it threads well."
You are right.
SPARC is Oracle's largest customer base for Oracle Databases.
That being said, it is highly unreasonable that Oracle is going to kill the chicken that lays the golden eggs after bring the chicken into the hen-house for protection!
I really don't understand how people can come to any other conclusion.
"....SPARC is Oracle's largest customer base for Oracle Databases....." But hp is Oracle's largest partner through the new Oracle installations on ProLiant. And ProLiant is also the largets MS SQL install partner, and that is where Oracle is looking to focus to try and stop MS SQL eating up the rest of their installed base. MS SQL has already overtaken Oracle as the most commonly installed database and Microsoft is making hungry looks at the rest of the Oracle stack. Oracle's big battleground is x64, not the high-end enterprise (where it sells the most new installation licences on hp Integrity anyway). MySQL offers Oracle an option for a a proper Oracle Lite based around the open source community's enthusiasm. Installed SPARC Slowaris Oracle is just an opportunity for migration licences as hp and IBM move to migrate all those Sun customers off that legacy SPARC base onto Power, Itanium, Xeon and Opteron. Oracle has more interest in keeping chummy with hp than trying to pursue a doomed processor line that dragged Sun down.
As to the other two Sunhsiners posting the ludicrous idea that Niagara is a good databse engine and that BTRFS is dead, you're not going to like these simple facts. Niagara is a completely pants as an Oracle engine, I know because I've tried it. Why else do you think Sun and FSC were forced to bring out the SPARC64-based M3000? And Fujitsu is too busy breaking up the old FSC, and doesn't have the cash to save SPARC. In truth, they seem to be struggling to pin a release date to the next gen SPARC64, so why they would want to lumber themselves with a bug-ridden and late product like Rock is beyond me. But then maybe that's because I don't wear one of those Sunshiner Blindfolds™ that seem so good at blocking out nasty things like reality.
And BTRFS has been adopted by the Linux community for the very good reasons that it had nothing to do with Sun, one of the most distrusted software vendors in the eyes of the community (kind of like a M$ Mini Me), and because it doesn't have NetApp filing a suit against it because it isn't a rip off of WAFL as even the Sun engineers admit ZFS is.
/only six more months of pointing and laughing at the Sunshiners before Oracle ties up the deal and starts cutting them off?
Ha, Ha, Ha... LOL... Matt, Matt, Matt... You're HP Enema is starting to spill out of your ears. HP as a preferred partner to Oracle is over. Solaris is again the preferred platform for Oracle, as espoused by Oracle.
Also, your comment that "Niagara is a completely pants as an Oracle engine, I know because I've tried it." Shows that you're either inept at benchmarking or you're a liar. I lean toward the latter, but the former seems just as likely. Niagara as a single or dual socket DB server has no equal. You obviously bought into the HP FUD over perf/core, but like them, forgot that it's about the system, not the core. Oracle runs very nicely on Niagara, and I know because I have done it and I do it. I run Oracle on Niagara at a very nice discount to HP and IBM, thank-you-very-much.
Matt Bryant comments, "But hp is Oracle's largest partner through the new Oracle installations on ProLiant."
Once Oracle may own their own Intel & AM line of processors, the question is whether this will remain the same... One would suspect that HP ProLiant sales may decrease in this category.
Matt Bryant continues, "And ProLiant is also the largets MS SQL install partner, and that is where Oracle is looking to focus to try and stop MS SQL eating up the rest of their installed base."
Since Oracle may own MySQL as well as the Oracle RDBMS soon... one might expect a two-fold assault against MS SQL.
Matt Bryant comments, "Niagara is a completely pants as an Oracle engine"
CoolThreads Servers seems to hold their own against HP Itanium.
For example, in the "SPECjAppServer2004 JOPS@Standard" benchmark, HP edges out a few T2+ platforms, requiring a whole lot more sockets.
Result - J2EE-Sockets - Oracle-Sockets
10519.43 - 24 - 20
09500.76 - 08 - 04
I would not suggest that a 4 Socket T2+ CoolThreads server as an Oracle engine that keeps up with a 20 socket Itanium server is unreasonable when "The Sun T5440 delivered 90% of the performance using 60% of the database licenses compared to the HP Superdome"
The CoolThreads processor is clearly cheaper in hardware purchase, software licenses, data center cooling, data center power consumption, rack space consumption, ethernet switch port consumption, and console port consumption.
If SUN CoolThreads makes a bad database server, then HP Itanium is a whole lot worse.
Of course, CoolThreads and Itanium are not designed for all workloads. I am sure there are other database exercising benchmarks that how Itanium to be reasonable.
RE: Anonymous Marketing Droid
"Ha, Ha, Ha... LOL..." Yeah, hope the laughter helps whilst you wait on your pink slip from Oracle. You know I'm going to have the last laugh as my future is not tied to Slowaris or SPARC, unlike yours.
"....HP as a preferred partner to Oracle is over. Solaris is again the preferred platform for Oracle, as espoused by Oracle....." Now you are making me laugh! Oracle can't afford to turn their backs on the vendor that is going to give them the most new licensing opportunities in the coming years. How do you think hp broke the old stranglehold that Sun had on Oracle from the dot-boom days? Hp simply became much more important, especially when they bought Compaq and the code that underlies the Oracle clustering technology. With IBM pushing DB2, hp is the natural partner for Oracle, especially as it means hp are less interested in pushing MS SQL further into the datacentre whilst the reltionship lasts. Like I said, Larry doesn't do fantasies.
"....Shows that you're either inept at benchmarking or you're a liar. I lean toward the latter, but the former seems just as likely....." Well, Sun were allowed to tune the Niagara kit for our shoot out, so are you accusing Sun of being inept? And they lost, because we tested real data with a real business app, and the Niagara choked. As a result, all our old SPARC Slowaris Oracle instances which Sun wanted to move to Niagara are now running on either hp-ux on Integrity or RedHat on ProLiant - the M-series couldn't compete on price or performance. Check the market figures, it's happening all over the place, because Sun haven't been able to provide what customers actually want at a competitive price for years.
".....You obviously bought into the HP FUD over perf/core, but like them, forgot that it's about the system, not the core...." As I have explained before, we don't buy anyone's FUD, we insist on benchmarking any major solution in our environment with real data, as that is the only way to get a real idea of how it will perform. The only one buying the FUD seems to be yourself, or how do you explain the many instances on the web of users moaning that Niagara just can't handle the heavy threads needed for real enterprise apps with Oracle? Or that it is just so ludicrously expensive compared to Xeon or Opteron kit?
"Oracle runs very nicely on Niagara, and I know because I have done it and I do it...." Yeah, I'm thinking back to your statement about liars. Either that or you're running a cutesy webapp with tiny threads, rather than a real enterprise database. Whatever, if it suits your business then enjoy, at least until your management make the strategic decision that SPARC is dead and they had better move off it to a platform that has a future. Is that the IBM and hp salesgrunts knocking on your CIO's door? You'd best go and try feed him some more of that Sunshine, just in-case his Sunshiner Blindfold™ isn't on quite as tightly as yours.
I see there is still a problem with reality out there in the Land of Sunshine.
"....Once Oracle may own their own Intel & AM line of processors, the question is whether this will remain the same... One would suspect that HP ProLiant sales may decrease in this category....." You fail to understand that customers are buying hp ProLiant over other vendors' x64 offerings because they prefer what they get from hp. Oracle, with no hardware experience, is just as unlikely to be able to take on ProLiant with Galaxy as Sun were. Oracle and Galaxy alone won't even stand a chance against Dell or IBM, let alone hp. The fact is Larry is going to take an axe to the Sun hardware bizz - Galaxy may survive as it allows Oracle to pitch their own inhouse verson of the hp Exedata and Database Machine offerings, but only to accounts where they aren't partnered with hp already. Expect some nice little chats between the very close hp and Oracle sales teams as to whom gets to play where. Is that the sound of your buble bursting I hear?
"....Since Oracle may own MySQL as well as the Oracle RDBMS soon... one might expect a two-fold assault against MS SQL...." What you should expect is Oracle to protect their enterprise DB Oracle offering by castrating MySQL into an Oracle Lite. Result - a fork in MySQL, as already announced by the community. Oracle will make some suport money from the non-Oracle MySQL, especially through InnoDB, but MS will still get the last laugh. Or didn't you notice that the original Sun purchase of MySQL did absolutely NOTHING to the rate at which MS SQL is ramping up?
As to more of your selective benchmark quotes, do you really want a repeat of the SAP benchmark spanking you let yourself in for last time? Do you really want me to go look at those results and find out, like the SAP ones, you are comparing new Niagara with five-year-old Itanium? I'm not going to waste my time with any benchmark you quote as I've already had Sun come round and sprout all their FUD directly, and had fun showing them up. If all you're going to do is repeat the same Sun salesguide then please don't bother - the market didn't belief it either, as shown by the fact that hp and IBM are gutting Slowaris accounts in the enterprise high-end.
An even better reason to ignore anything you Sunshiners say about Niagara is that Oracle have said NOTHING about any future for it. MySQL - rapid response to negative community comments, attempt at reassurance - so tick. Java - tick. Slowaris - tick, kinda, still no details. Any Sun hardware? Big, long silence. Nothing, nada, nil point! Galaxy as storage is just guessing by the analysts, nobody outside of the Land of Sunshine even thinks Rock, T3 or the rest of the StorageTek side have a chance in Oracle hands. Carry on imagining that Rock is "too far down the line" to be cancelled, forget that even the SPARC fanclub canned UltraSPARC V at a later stage when it became obvious it was too little too late, and don't for a second remember that the SPARC fanclub are not even in control any more. You Sunshiners can squeal all you like, but the analysts that the market listens to are already looking at whom is going to buy up the bits Oracle doesn't want, and there doesn't seem to be any buyers lining up for anything SPARC. Did you forget, hp already had the chance and said "no", IBM dumped them, and Fujitsu have already said they don't have the cash spare? There's no-one left, even your latest fantasy dribbling of Apple buying up Sun have proven just as stupid as your idea that FSC would.
/must record some of these Sunshiner sulks, there's only going to be about another six months of this comedy to point and laugh at!
Blame it on hp - provisioning LUNs with CommandView is so quick I actually had the time spare to go and look at the SPECjAppServer2004 results. And what grabs your eye first? Top result is an hp BL870c blades cluster with Oracle WebLogic Server Standard Edition Release 10.3, on hp-ux 11i v3, scoring 26,655 JOPS. Yes, hp's Itanium-based blades scoring higher than ANY Sun server solution, from blades right through to M9000. The highest Sun result published is just 9,501 JOPS for a cluster of four T5140 Niagara servers, again using Oracle WebLogic Server Standard Edition Release 10.3. Sun have always had problems scaling Niagara, so whats the excuse for the lack of an M-series result even close to the BL870c cluster? Isn't the M9000 man enough for the job? Guess not.
Talking of blades, shall we go look for results for the latest T2/T2+ blades from Sun, released back in October 2008, the T6340 and T6320? Direct competitors to the BL870c, you'd expect Sun to have lots of benchmarks ready to fight their corner, right? Erm.... no! No SPECjAppServer2004 result. CINT2006? No published result. CFP2006? Nyet, nien, non! Doesn't look like Sun thought there was much point in benchmarking their new blades seeing as they have such a tiny share of the blades market compared to hp, IBM or Dell. Makes you wonder why they even bothered making them seeing as they must have known back then the company was up for sale and anything SPARC had zero future. I guess they just hoped to sucker in some more Sunshiners to get a few more sales in whilst they looked for a buyer.
In fact, although there are some SPEC results for the T5xxx servers, there are no published T2/T2+ results for CINT2006 or CFP2006, probably as these highlight individual core performance. Sun still trying to hide the dreadful performance of their wheiner cores?
/waiting for Sunshiner meltdown in 3... 2.... 1.....
Who is Novatose? I don't see him posting anywhere.
Let me get this straight...
An Itanium score of 2.8x faster than a SPARC score... at what cost?
4x as many Itanium J2EE nodes than SPARC nodes
8x as many itanium J2EE sockets than SPARC sockets
16x as many of Itanium database sockets than SPARC sockets
There is NOTHING to be proud of here, Matt.
You say SUN Niagra processors have a problem scaling when your benchmarks demonstrate that Itanium can't scale?
Those were not the even fastest T2+ chips available from Sun today, on the J2EE servers.
I think every business wants to know that they can run a database 2.8x faster than a 5U high SPARC by adding 16x the number of sockets and a couple racks of HP Itanium equipment!
What a comedy!
Do the HP Itanium systems cost 10x as many Euros than the SUN SPARC systems? or is it 100x the Euros for that little 2.8x speed bump?
The poor HP Itanium database server needed 7 dual-ported 1GigE Ethernet cards while the SUN just needed it's 10GigE Ethernet card. A little deficient in the I/O bus speed? Can't dig a 10GigE from a closet? Need a SUN Ethernet card?
HA HA HA HA!
You showed the world how HP Itanium is completely out-classed by a slow threaded SUN T2 SPARC... not to mention, needs a severe I/O upgrade!
Matt Bryant compares an 5U high sun box - JUST WAIT TILL LATER THIS YEAR WHEN HP HAS 4 CORE ITANIUM'S - WE WILL ONLY NEED 8x AS MANY MANY SOCKETS TO BE 2X FASTER! WE'LL HAVE 1 FEWER RACK THAN BEFORE! WE ONLY NEED 7X THE NIC CARDS! IT ONLY COSTS 10X MORE! WE'RE WINNING!
HA HA HA HA!
Matt Bryant, at least 1 year behind in technology, at least 4 years behind in school, can't spell Solaris, can't get a job, can only enjoy fantasies on-line, and talks to non-existent people.
Just the kind of guy who no one listens to.
L O S E R ! ! !
Matt Bryant posts, "shall we go look for results for the latest T2/T2+ blades from Sun, released back in October 2008, the T6340 and T6320? Direct competitors to the BL870c, you'd expect Sun to have lots of benchmarks ready to fight their corner, right?"
The T2 and T2+ processor scales similarly in multiple SMP systems. How does a T2/T2+ compare to a BL870c in the SPECjAppServer2004 benchmark?
JOPS Nodes Model
10519 12 HP-UX Integrity BL860c
09501 04 Sun SPARC Enterprise T5140
The HP Itanium server requires 300% more nodes for a 10% performance gain over the Sun SPARC models. (An equivalent SPARC T2+ blade would be the T6340 blade module.)
Matt Bryant suggests, "Sun have always had problems scaling Niagara"
Using the SPECjAppServer2004 benchmark you suggested, to compare Itanium to SPARC Niagra, we can extrapolate the scaling factors on a dual-socket node.
Your benchmarks suggest HP "had problems scaling" Itanium in contrast to SUN with SPARC Niagra.
A friend at Oracle just reminded me that Oracle chose hp Integrity servers when it moved off Exchange to their own Collaboration Suite for all Oracle email users years ago. Their back office is all ProLiant, much of it on RedHat. Looks like Sun's own IT team had better get on with some cross-training!
"....An Itanium score of 2.8x faster than a SPARC score... at what cost?...." Well, just for you Sunshiners, I did try using the Sun online shop to price up an equivalent T5140 spec, only the online tool doesn't let you put more than 32GB of over-priced RAM into the SPARC box, and I ran out of PCIe slots and couldn't put in an internal RAID card for the boot disks way before I got to matching the number of fibre and Ethernet ports out of the BL870c. And then, there's the cost of the additional switches and mangement tools that you'd have to throw in to match the hp blades (I tried doing the same with the Sun baldes but you just can't get to an equivalent spec).
"....There is NOTHING to be proud of here, Matt...." No, nothing like having the fastest benchmark for a software solution you Sunshiners keep telling us is a key technology. Oh, hold on a sec - didn't Sun used to push these benchmarks like crazy? Could it be that since they have fallen off the performance map they just can't be bothered in even trying to benchmark their solutions? Did Ponytail lay off the testing team early?
"....Do the HP Itanium systems cost 10x as many Euros than the SUN SPARC systems? or is it 100x the Euros for that little 2.8x speed bump?...." Strange that hp seem so capable of supplying better systems at less cost to the customers then. Did you use the same warped calculator you use for all those Sun performance claims to calculate the costings there? Lots of research presented very neatly - not! Care to post some pricings to back up your wild claims (whilst your Sun rep can still give you prices)? I'm betting not.
"....The poor HP Itanium database server needed 7 dual-ported 1GigE Ethernet cards while the SUN just needed it's 10GigE Ethernet card. A little deficient in the I/O bus speed? Can't dig a 10GigE from a closet? Need a SUN Ethernet card?...." Lol, if you knew anything about other vendor's kit you'd know the Virtual Connect modules in the C-class blade chassis come with 10GbE ports. Looks like you know as little about the competition as you do about what customers actually want.
".....WE'RE WINNING!...." A quote that says it all about the deluded Sunshier mentality. How can losing money for years, losing marketshare for years, being late to every market with every product, being rejected for buyout by the two leading hardware vendors, going from a $200bn company to less than $6bn in real cash, refusing to let customers benchmark your kit because you know customers will then see how poorly it performs in real world tests, and finally being swallowed up at peanut value by a software company, class as winning? Sun is dead. SPARC is dead. Go find another cult.
RE: Matt Bryant: still looking for mr. goodbenchmark
Novatose, why do you bother posting as AC and then immediately after as David Halko? Your syntax, writing "style" and repeated innaccuracies and FUD are so obvious. Have they fired the rest of the marketeering droids already?
"....10519 12 HP-UX Integrity BL860c.... 09501 04 Sun SPARC Enterprise T5140...." And still pointing to a rack server rather than the new Sun CMT blade, and then comparing it to a different hp blade. Admit it, it's because you couldn't find any Sun benchmarks for the CMT blades, Sun have avoided them as they know the awful chassis design only compounds the poor CMT blade performance.
"....An equivalent SPARC T2+ blade would be the T6340 blade module....." Nope, not equivalent, they are different products. I would agree that both can't offer the performance or flexibility of options as the BL870c, but as to how the T6340 and T5140 compare you can't state as you don't know. Sun are very keen to keep any such comparison under wraps as they know how poorly their whole third failed attempts at blades do in comparison with hp blades.
"....Using the SPECjAppServer2004 benchmark you suggested, to compare Itanium to SPARC Niagra, we can extrapolate the scaling factors on a dual-socket node...." Hilarious! You really don't have any idea how clusters work, do you? First off, you need reliable clustering and scaleable software and hardware - Sun can't post results for large T2/T2+ clusters becasue they know they scale clusters poorly. Secondly, you have the additional traffic in clusters that degrades performance as you add nodes - adding a fifth node does not guarantee a 25% increase in performance over four nodes, and it gets worse as you grow the cluster. So any extraploation is just the usual Sunshiner fantasy groping in the dark as Sun won't post large cluster results. Until they do, you can extrapolate all you like and no-one will believe you.
/maybe swine flu struck early, we just didn't notice the symptoms included delusional belief in Ponytail?
I see IBM as the loser in this. Oracle will likely continue to partner with HP. They would be insane not to. Oracle likes to get paid for commercial use of their products, but I've found them very reasonable to work with. They provided me with an evaluation copy of their software to prototype a recent project and even set up a conference with four of their top engineers when we ran into problems. Oracle also gives away free copies of their enterprise database for noncommercial use. I've worked with them since 1980 and while some of their HR practices may be unduly harsh, if you're a customer they're generally good to work with in my experience. It makes sense to let HP take the hit for R&D on the hardware side or at least acquire it from Intel and Hitachi. I think Oracle will spin off or kill most of Sun's business except for Java. Since WebSphere, IBM's cash cow, is dependent on Java, this kind of puts Oracle in the same position as Microsoft with Windows and Office. They can create a moving target hard for others to hit while skirting monopoly laws.
Matt Bryant posts, "there are no published T2/T2+ results for CINT2006 or CFP2006... you couldn't find any Sun benchmarks for the CMT blades"
There have been "CINT2006 Rates" and "CFP2006 Rates" benchmarks (that you could not find) published for the CMT blades since February 2008. The CMT blade benchmarks used 1.4GHz processors, vs the SPARC T2 1.1GHz processors that out-performed the Itanium processors 2376 JOPS/node to 876 JOPS/node.
Matt Bryant speculates, "Sun have avoided them as they know the awful chassis design only compounds the poor CMT blade performance"
Apparently, Sun did not avoid the benchmarks, as you wrongly speculated before.
Now, for your Sun chassis speculation. There was a SUN blade chassis benchmarked (your SPECjAppServer2004 benchmark) in July 2007 with 1 chassis SUN 10x 1socket SPARC T1 processor blades, that provided slightly slower performance to the 2 chassis HP 12x 2 socket Itanium processor blade at a substantially lower cost.
HP needed to over double the Itanium processors and double the HP Blade chassis count over the single Sun blade chassis filled with single socket T1 for a slight performance boost. Your cited benchmark demonstrates inferior HP chassis design, or severely inferior processor performance, for this workload.
You clearly did not know: SUN CMT T2 blades had benchmarks, SUN CMT T2 blades scored comparable benchmarks to rack units, a single T1 SUN blade chassis scored similar benchmarks to double the number of HP Itanium chassis, and lower clock rate CMT T2 processors out-performed the Itanium by over a factor of 2. You were not able to cite evidence for your incorrect information with evidence.
A reasonable reader would doubt your continued speculation concerning blade chassis design without any citations.
David Halko posts, "....An equivalent SPARC T2+ blade would be the T6340 blade module....."
Matt Bryant posts, "Nope, not equivalent, they are different products... how the T6340 and T5140 compare you can't state as you don't know"
One can compare when understanding: system architecture (standard interfaces are leveraged without proprietary bus bottlenecks), common shared IC's between blades & racks, and existing benchmarks (that you did not know existed.)
The Sun Blades pass through an industry standard buses (like PCIe) right to the back of the chassis, just like the rack server.
The SUN SPARC T2+ blades benchmarked equivalent (slightly faster) performance than the rack mount server, which runs at the equivalent clock rate. Using 30% faster clock rate CMT SPARC T2+ blades would make the overall performance of all nodes higher.
Matt Bryant writes, "Sun can't post results for large T2/T2+ clusters becasue they know they scale clusters poorly."
The last few times you speculated without evidence, the benchmarks that you could not find demonstrated you were wrong about the SUN blade chassis. Architecture of the Sun blade chassis tells the world a different story than yours.
Do you have any reference to substantiate yet another nefarious claim?
Matt Bryant writes, "Secondly, you have the additional traffic in clusters that degrades performance as you add nodes - adding a fifth node does not guarantee a 25% increase in performance over four nodes, and it gets worse as you grow the cluster."
Moving from 4-5 nodes will be pretty close. A single SPARC node guarantees vastly superior performance boost to Itanium under the workload, using your suggested benchmarks.
Using your argument, no one would want to buy an HP Itanium instead of the SUN T2 SPARC for this workload, since you can get the same performance with far fewer SPARC node, or even superior performance with an uplifted 30% clock rate on T2 SPARC, since adding 3x as many Itanium node demonstrated such diminishing returns.
".....There have been "CINT2006 Rates" and "CFP2006 Rates" benchmarks..." Ha! Did I ask for Rate benchmarks? Nope, I asked for the CINT2006 and CFP2006, the tests that shows the single-threaded ability of a CPU. The Rate tests allow you to push lots of wheiner threads through many cores, which is what Niagara was desinged to do even though it has no relevance to the Oracle market. Of course, there is no way Sun want to show the ordinary CINT2006 or CFP2006 figures as they know any CMT CPU is pants at the type of Oracle app the installed SPARC Slowaris base is actually serving. Hence their failure to get SPARC customers to shift from old SPARC to Niagara. Those customers that remained faithful to Sun and waited on the promised Rock now have two options - buy a Fujitsu M-series server and worry about the future of SPARC64, or move off onto a real enterprise CPU with a future, like Power or Itanium. Some may be able to move to Slowaris on x86, but after Sun has let them down so badly, and Oracle have not reassured them, it will be Opteron and Xeon kit from other vendors, most probably the market-leading ProLiant. Niagara is going to whither and die, probably as another SPARC product "open-sourced" to a market that has zero interest in it. No other vendor has stepped up to fab T1, T2 or T2+ chips, and no-one has shown any interest in making a T3 (including Oracle).
The whole of your post is a long-winded and deceitful attempt to try and convince people Niagara doesn't have a problem with single-threaded performance or clustering problems, and you avoid both issues by going off into areas that are either nothing to do with the kit in question, or just plain deceitful. "....One can compare when understanding: system architecture (standard interfaces are leveraged without proprietary bus bottlenecks), common shared IC's between blades & racks, and existing benchmarks (that you did not know existed.)...." No you can't, because the whole backplane design is completely different between the Sun blades and the rack servers, the components are different, and only a complete ignoramous or a liar would try and pretend the difference would have no impact on performance. The hp DL380 racked server and the BL460c blade post different results because they have different backplanes and different board designs, even thought they use the same "industry standard interfaces" and are two-socket Xeon servers. In fact, the new hp G6 range has a whole number of two-socket designs using the same industry standard interfaces, are you going to pretend they all perform exactly the same?
All you have shown is you either don't understand architecture or don't want people to see the difference. Either way, since Sun haven't released the CINT2006 or CFP2006 benchamarks (I wonder why) or benched as large a cluster, you can extrapolate, bluster and lie all you like, no-one will take your word for it. In fact, the vast majority of customers that have been burnt by Sun's performance promises in the past will just flat out not believe you. Why do you think we moved to a policy of shoot-outs? It wasn't becasue we found that hp or IBM gave us unrelaible estimates, it was because we found Sun's increasingly extravagant claims and wild FUD to be unsubstantiated.
Moving to shoot-outs does seem to have prolonged many purchase cycles, but the payback is we get to spot a lot of issues (app, stack, storage and server, or just with our own processes) long before we hit production. It has also massively reduced the amount of FUD and promises we get from the vendors, as they now know we will be putting their statements to the test. Anyone even tempted to swallow Novatose's male bovine manure, I would recommend you do a real test before you hand over your money, as nobody even knows if there will be a "Sun" hardware bizz to complain to in a year or so.
/point, laugh, enjoy the Sunshiner deathrows!
Matt Bryant posts, "there are no published T2/T2+ results for CINT2006 or CFP2006"
Matt Bryant posts, "you couldn't find any Sun benchmarks for the CMT blades"
Matt Bryant posts, "Did I ask for Rate benchmarks? Nope"
You are incorrect again. You posted on Thursday 30th April 2009 11:11 GMT "any Sun benchmarks for CMT blades" - a simple find command on the page shows that. Also, I identified a 6000 chassis with CMT blades with non-Rate benchmarks. Clearly, the benchmarks identified qualify as "any".
Matt Bryant posts, "failure to get SPARC customers to shift from old SPARC to Niagara"
You are incorrect again. Q2 2009 results show increasing uptake on CoolThreads every quarter as well as year-on-year.
Matt Bryant posts, "try and convince people Niagara doesn't have a problem with single-threaded performance or clustering problems"
You are incorrect again. I never discussed single-threaded performance. I also posted a clustered performance metric, showing a single Sun blade chassis of clustered blades being close to performance as 2 HP chassis combined with over 2x the Itanium sockets.
Matt Bryant posts, "Why do you think we moved to a policy of shoot-outs?"
You claimed to have posted elsewhere to have already replaced all of your Sun equipment, so is this comment about this claimed policy incorrect? (The term "incorrect" here may be generous.)
Matt Bryant posts, "we found Sun's increasingly extravagant claims and wild FUD to be unsubstantiated."
I have not seen you post substantiated "extravagant claims" with "wild FUD" from Sun.
You frequently use FUD with unsubstantiated claims, many of which have been demonstrated incorrect post after post.
Your behavior appears to be a classic case of psychological projection.
"....You are incorrect again..... Clearly, the benchmarks identified qualify as "any"...." Nope, I specifically asked you to find the CINT and CFP benchmarks for the T6340 and T6320. You have prevaricated, wandered off on sidetracks and just plain lied, but still not supplied those benchmarks for what is supposed to be Sun's great white hope for blades.
"....You are incorrect again. Q2 2009 results show increasing uptake on CoolThreads every quarter as well as year-on-year...." All the Q9 results showed was Sun making a loss, again, as it has for years. All Niagara has ever done is cannibalise a few of the remaining UltraSPARC webservers, Oracle's own figures show a decline in Oracle on Solaris licence sales, which proves your whole line is just waffle.
"...You are incorrect again. I never discussed single-threaded performance...." Lol, no, you Sunshiners never want to discuss single-threaded performance and Niagara as you know you get whipped every time! Which is why I asked specifically asked about CINT and CFP, and why you deliberately avoided them and went for the rates results instead.
"...I have not seen you post substantiated "extravagant claims" with "wild FUD" from Sun...." I was referring to previous Sun sales meetings, or are you admitting to a Sun relationship?
"....You frequently use FUD with unsubstantiated claims, many of which have been demonstrated incorrect post after post...." Really? Please point to one "claim" or "FUD" you have managed to disprove. Do you deny Sun hasn't made a profit for years? Or do you claim it is a lie that Sun wnet from a $200bn company to less than $4bn by market cap in less than a decade? Or that hp-ux on Integrity and AIX on Power are taking marketshare from SPARC Slowaris in the enterprise high-end (you know, the profitable bit where all those nice big Oracle instances live)? Oh, and please do try some more selective benchmark figures as they really make me laugh.
"....Your behavior appears to be a classic case of psychological projection...." Your behaviour is just the typical Sunshiner, unable to cope with the fact that Sun has been beaten into submission by hp, IBM, and Dell. That makes all you Sunshiners that swanned about for years saying only Slowaris on SPARC wrong. For years you gave us that baloney, and it was proven to be just that. And now you sprout junk about how Oracle is just going to keep on with al those Sun products that dragged Sun down and just expect us to believe you? I'd like to say if you weren't so comic you'd be pathetic, but you're both.
Wow Matt, you've excelled in this little posting frenzy, your posts are longer than ever. You either a very bored sales bod or else this whole Oracle/Sun aquisition has seriously rattled your cage. At least you've stopped posting about the hardware selloff, things are slowly sinking in.
I've never seen so many toxic posts since this kicked off (periodically I go fishing for you, just to see if you'll bite but your in a rabid snapping fury these days)
Your behaviour would almost be funny if it wasn't quite so sad that your so dammed serious about your HP loving! Just assure me you didn't re-mortgage to buy shares in HP recently? Did you? Tell me you didn't use the kids savings as well?
Any other reasons for the 1-2000 gibberish word replies that act to prop up HP position in life?
PS: A little hint Matt, HP "was" a favoured partner (note : past tense), I hope your bright enough to see that changed now with statements from Oracle such as "Solaris is the number one Unix"
Repeat after me Matt, "Solaris is the number one Unix, Solaris is the NUMBER ONE UNIX!"
Now where is that old post where you agreed with the T2 performance being better than AMD & Intel boxes, let me go-a-hunting...... Hee heee heeeee....... :-)
"Any Sun hardware? Big, long silence. Nothing, nada, nil point!"
You lack the ability to read, I know, but even you could have read this from Oracle's FAQ on the deal:
"Oracle plans to grow the Sun hardware business after the closing, protecting Sun customers’ investments and ensuring the long-term viability of Sun products."
So, Oracle has said nothing about Sun Hardware? Nothing, nada, nil? Hmmm. You out do yourself Bryant.
Even the IBM PR machine IDEAS International thinks that Oracle is moving into the HW business for real "Oracle Really Entering the Server Hardware Business?"
RE: Troll Bryant.
<Looking for any form of technical argument....> Oh, there is no technical argument, just more blathering and insults. A complete lack of technical facts or discussion? You must work for Sun sales.
".....Repeat after me Matt,..." Lol, I know you Sunshiners try that indoctrination stuff - "don't think, don't reason, just repeat after me" - but I've already had far better Sun salesbods than you try it and they failed, because we made them try and prove their claims, which they couldn't. Slowaris is only number one in profit and marketshare losses, and both will only accellerate as Larry guts Sun.
RE: Re: MB rant
"...."Oracle plans to grow the Sun hardware business after the closing, protecting Sun customers’ investments and ensuring the long-term viability of Sun products."...." <Yawn> Yeah, and the bit that expressly ties Oracle to a SPARC roadmap, to a future for Rock or T3, or even a Slowaris roadmap is....? Oh, there isn't one! Now, repeat what TPM put in his Slowaris 11 article: "So until this deal is done - or undone - it is hard to say when any Sun product, be it hardware or software, will appear."
Until Oracle release some roadmaps, which won't happen until after the purchase completes (and that's if M$ and IBM don't have fun throwing up some objections), nobody has any commitment to ANY Sun product. Until then, Sun's whole product range has just become vapourware (well, the few bits that weren't already!). Anyone considering buying any Sun hardware under that uncertainty is, in my professional opinion, frankly, a sucker.
/making the most of the Sunshiner comedy whilst it lasts!
Biting the hand that feeds IT © 1998–2021