Gates touts Microsoft's supercomputing move

Microsoft Chairman Bill Gates on Tuesday announced the company's foray into the world of supercomputing, though its first operating system for computer clusters remains in beta testing.

Speaking at a supercomputing conference in Seattle, Gates announced that the company has reached the Beta 2 stage for its Windows Compute Cluster Server 2003. The product consists of both a cluster-optimized version of Windows Server 2003 as well as software for job scheduling and other tasks. It is scheduled for release in final form in the first half of next year.

"Technical computing is crucial to the many discoveries that impact our quality of life--from making safer, more efficient cars and airplanes to addressing global health issues and environmental changes," Gates said in a statement. "Moreover, most sciences are becoming computational sciences, which is why advanced computing capabilities need to be seamlessly integrated into the end-to-end scientific process."

Separately, Microsoft also announced that the Compute Cluster Server and several other upcoming server software releases will work only with 64-bit processors. Such chips, which include Intel's 64-bit Xeons and Advanced Micro Devices' Opteron, are becoming the norm on servers, and 64-bit processors are making their way onto desktop machines.

"We are making big bets on 64-bit technology and working closely with our industry partners to enable a smooth transition for customers, so they can begin to realize the benefits of mainstream 64-bit computing," Microsoft Senior Vice President Bob Muglia said in a statement.

Other titles that will be only 64-bit include the next version of Exchange, as well as the upcoming "Centro" midmarket server and the Longhorn version of Small Business Server. Microsoft is planning a version of Longhorn Server that will work on 32-bit chips, though it expects the first major update to Longhorn Server, Longhorn Server R2, to be exclusively 64-bit capable.

Academic institutions and some industrial customers have been combining clusters of standard Intel- and AMD-based servers for some time. But Microsoft says it has seen a shift where such products expand beyond a niche market into more and more businesses. Microsoft is pitching its tools as on par with the performance of Linux. The company also claims its tools are easier to manage and integrate with the rest of a corporate computing environment.

"HPC (high performance computing) is starting to broaden out," said Kyril Faenov, director of Microsoft's HPC unit. "What that leads to is demand, on behalf of customers, to really provide this raw power in a way that is easier to consume and easier to integrate into what they are already doing."

CNET News.com reported in May 2004 that Microsoft was planning a high-performance computing version of Windows, a move later confirmed by Microsoft.

In March, a Microsoft engineer said the company hoped to have a product out by fall, though that turned out to be a beta. The new beta version will be public, unlike the one released at September's Professional Developer Conference in Los Angeles, which was limited to about 1,600 testers.

In addition to announcing the new beta, Microsoft is touting the support it is receiving from hardware makers. The company will outline 19 key applications, from other software makers, which will run on the new version of Windows.

And, Microsoft also announced investments in 10 high-performance computing institutes that will serve as early customers and help the company determine where to go with the cluster-software effort.

[heystoopid]

too late

too late linux/unix holds the majority market share in this market!

[Bobman]

Never too late

Just look at how much AMD is cutting into Intel's market share.

[folsco]

Never . . .

Also, look at how ".Net" is eating up "Java" market share! Can we really say it is too late for anything now? I don't think so.

[just_some_guy]

When hell freezes over

It will be too late when progress grinds to a halt and the knowledge of mankind ceases to advance.

You might as well say it is too late for Linux, or any operating system other than Windows, to become a dominant consumer platform - so they might as well stop trying and just accept Windows for the rest of eternity.

[SparXXXie]

too late?

Linux and Unix also holds the market share for servers, but you don't see M$ backing off....

[MahRain]

Xgrid, anyone?

Apple already supplies a piece of software called Xgrid with it's computers to use a 'Seti@home' like environment on regular computer networks, to turn it into a computing grid. All you have to do is tick a box in the network preferences.

Oh I bet Microsoft will find a few high-profile customers for their OS, but I think Apple's solution is far more interesting.

good point...

cluster is such a vague term... and a real ***** to market..

on my xbox I want a partitioned failsafe software host with automated job distribution... as for storage... I read about some native scsi over sata.. its supposed to even would in a multipath setup....

[kakman1]

Be afraid be very afraid

I am very afraid who is going to buy this. I'm sorry Bill, but your domestic and business software doesn't inspire me with any great confidence. Good enough isn't good enough in this field. It has to be better. Much better. I'm praying that the military, NASA, the weather agencies, the pharmaceutical companies, etcetera, give this product a pass. Anyone who elects to buy this best have their lawyers write a EULA agreement before you sign the dotted line, so Microsoft wont be able to wriggle of the hook when things start going awry.

[shadowself]

Yes, be afraid. Be very afraid.

Once the Microsoft Windows system for HPC comes out (probably based upon XP or maybe even some earlier version) many users will blindly go for this. Some possible examples are the U.S. Navy and the U.S. Air Force. The U.S. Navy a few years ago had their CIO decree that all systems whenever possible will run some version of Windows (even running critical systems on ships!), and the Navy has never officially backed down from this position. The U.S. Air Force had a similar decree, but to a lesser extent. Thus if they buy HPC systems 3+ years from now many of those systems may be running some variant of Windows. I can just see it now, BSOD while running critical simulations of the missile defense system.

Thankfully, the U.S. DOE has not made a similar decree nor has the DSWA (formerly DNA) and they run more of the nuclear simulations. So while I'm a bit afraid, I'm not completely terrified.

[SparXXXie]

read the print before you comment

"And, Microsoft also announced investments in 10 high-performance computing institutes that will serve as early customers and help the company determine where to go with the cluster-software effort."

[Norseman]

Does it seem to anyone else...

...that Microsoft is all of a sudden going off in about a thousand
different directions? Where's the focus? Maybe they should get
one thing done well before trying 999 other things.

[just_some_guy]

Reminds me of another company...

Google.

[jbob70]

$40,000,000,000...

What would you do with $40 billion dollars?

[open-mind]

How To Compete

Usually when MS starts competing in a new market like this, they usually find a way to leverage their ubiquity/dominance in some other market.

But I dont' see how they can do that here. It seems like their product will need to be better than the alternatives.

Maybe if they could tranparently harness the huge amounts of unused compute-power on user desktops in the typical Windows-centric office network?

I don't see how else they're going to compete.

[Bill Dautrive]

Exactly!

"It seems like their product will need to be better than the alternatives."

Something MS has never done, although there are a small handful of arguable examples.

[System Tyrant]

Enterprise Marketing.

Blade Servers would be one area. How about clustering those 25 PC's they were going to through out because they were two old.

Don't get me wrong. Most companies are trying to lower the amount of hardware they have, but research labs are another story. Internet Service Providers could probably use clustering to lower cost and use computers longer.

I have no doubt that Microsoft will break into this area and they will find a way to market their products to anybody from home users to enterprise users.

[Lolo Gecko]

I'm glad ...

I don't own any Microsoft :)

right on!

clusters are such a ***** to configure... I hope I can evaluate the new software... I dont have an action pack though... I would be open to barter for one...

I just hope I wont be forced to study for some superficial exam or something... its my number one inhibition towards Microsoft products...

[Thomas, David]

Faster Crap

Knowledgable people won't buy into this product from Microsoft. Current MS corporate users may, only because they simply don't know better.

If MS has so many problems with their current OSs (not withstanding so many ignorant souls who have never really had the opportunity to truly understand they have been conditioned with mediocrity), how in the world do they think anyone would want a faster cluster--uck?

My guess is, their OS is so un-ruly, and bloated, they believe it will "sing", only on massive horsepower. Faster crap, is just faster crap.

The people who actually use grids, clusters, and any other derivations to get their supercomputing needs, are laughing at this. However, they are not the majority of computer users, so it is probably MS will sell to some. But in reality, if anyone who is serious about needing that kind of power, will NEVER look at MS for that solution.

[R. U. Sirius]

Does Microsoft Understand...

...that grid clusters have to be fault tolerant? Do they understand that just one point of failure can bring down the grid unless it is iron clad? Do they understand how spyware, malware, and viri could spread exponetially faster through a grid?

I am really beginning to despise Microsoft. When their focus was the desktop OS and Office, it was fine. But now they want to own everything with software based on a philosophy of "just good enough."

[Lolo Gecko]

this gives new meaning to 'Clueless in Seattle'.

:)