Windows Compute Cluster Server 2003 Released

grammar fascist writes "According to an Information Week article, on Friday Microsoft released Windows Compute Cluster Server 2003." From the article: "The software is Microsoft's first to run parallel HPC applications aimed at users working on complex computations... 'High-performance computing technology holds great potential for expanding opportunities... but until now it has been too expensive and too difficult for many people to use effectively,' said Bob Muglia, senior vice president of [Microsoft's] Server and Tools Business unit, in a statement."

Comments

Ok, I'll be the first

[Weaselmancer]

Imagine a Beowulf cluster of those.

Re:Ok, I'll be the first

[Ethan Allison]

But will it run Linux? (tm)

Re:Ok, I'll be the first

[Anonymous Coward]

How long will take to compute a blue screen? is it faster or does it take the same time? Will it be able to process more pop-ups from spyware faster? MS Cluster Fuck 2003-- MS Clustos

Re:Ok, I'll be the first

[Anonymous Coward]

Sounds like it'd be a clusterfuck.

Re:Ok, I'll be the first

[hdparm]

Not only that - it can predict the release date for vista with the error margin of just 2 years.

Re:Ok, I'll be the first

[nephridium]

I felt a great disturbance in the force, as if millions of CPUs cryed out in terror and suddenly silenced. I fear something terrible (but predictable) must have happened.

Re:Ok, I'll be the first

[nephridium]

Imagine a Beowulf cluster of those.

You forgot to add "The horror! The horror!" (with Marlon Brando's voice from Apocalyps Now)

Re:Ok, I'll be the first

[Scarletdown]

Or more simply...

*** Wilhelm Scream ***

Re:Ok, I'll be the first

[Anonymous Coward]

Imagine a Beowulf cluster of those.

Imagine patching Beowulf clusters every month.

Re:Ok, I'll be the first

[GotenXiao]

Yeah, you might just get the same performance as a beowulf cluster of 386 Linux machines ;)

Re:Ok, I'll be the first

[sl4shd0rk]

> Imagine a Beowulf cluster of those.

Bluescreening or running adaware?

Re:Ok, I'll be the first

[Anonymous Coward]

Imagine a botnet cluster of those.

When 99% of the worlds supercomputers runs this... MAY GOD BE WITH US

Too expensive my arse

[Umbral Blot]

Hasn't running "parallel HPC applications aimed at users working on complex computations" traditionally been done under Unix, and Linux as well. Seeing how Linux is free it's hard to see how "it has been too expensive", or "too difficult" (since unlike your home user the people running these systems are rocket scientists, I am sure a little command line use doesn't stump them).

Re:Too expensive my arse

[jbplou]

Well the hardware to setup these systems up is one of the major problems for the price. Most researchers have limited funding. I doubt using WIndows is going to help the price problem at all though.

Re:Too expensive my arse

[jbrader]

As I understand it this sort of thing can be done on just about any kind of computer. And at every university I've ever been to there's usually stacks of old pcs laying around.

Re:Too expensive my arse

[grammar fascist]

I doubt using WIndows is going to help the price problem at all though.

If they have to use Vista it'll probably exacerbate the price problem.

Re:Too expensive my arse

[supun]

Your response is the same as mine.

When I worked a Motorola, and was part of their LUG, one of the members was talking about a Beowulf cluster they made. Like bad management, they ordered a bunch of desktop PC they couldn't use, and no one authorized their return. So they sat around in unopened boxes until his team decided to make a Beowulf cluster so they could model the electron flow around traces in an 8 layer circuit board before they had them actually pressed.

Each prototype board cost around $10,000 to create. And after that you have to test to make sure the electron field, around a trace, does not affect another trace. Manually it took a long time and is prone to errors. So if there is a problem, it's another $10,000, and another, until you get it correct.

With this Beowulf cluster they could model the electron flow around a trace and then only make one prototype, saving a ton of money and time. And this was all done with an ISO off the net and a bunch of forgotten computers.

Re:Too expensive my arse

[Bananatree3]

[lol] Yep, you'll need half the cluster just to run the Aero interface! [/lol]

Re:Too expensive my arse

[geoff lane]

You can build an HPC from random PCs but it will be crap because the PC to PC interconnects will be too slow. Real HPC needs highspeed, low latency internal interconnects and these are expensive. But I fail to see how paying a "Windows" tax will make matters cheaper, or easier.

Re:Too expensive my arse

[0racle]

Not to many are using Fedora or Slackware on some white box with parts from Best Buy to do HPC. They have been altered to specifically run on hardware that was made specifically for this, and even then management of it is not exactly simple. Not that I believe that 2003 Server will suddenly change that but just using Linux somewhere does not automatically make it the cheapest way.

And I believe the correct answer to your question is Traditionally it has been done by tuned versions of commercial Unices which added to the base cost of the OS over and above the very expensive custom built hardware. Recently Linux has become able to do many of these tasks by similarly being modified at a significant cost running on the same expensive custom hardware. The recent HPC installation using mostly off the shelf parts (they didn't use Ethernet) was the one at Virginia Tech and that ran OS X, not Linux.

Re:Too expensive my arse

[kylegordon]

You can still have real HPC with slow interconnects. It all depends on the application for the HPC. If your data has a high scatter rate that requires large amounts of data transfer all the time, then you need fast interconnects. On the other hand, if your data can be sent off to a node to be crunched on for 2 hours, then a bog standard gigabit ethernet interconnect will do you just fine.

Re:Too expensive my arse

[joib]

Not to many are using Fedora or Slackware on some white box with parts from Best Buy to do HPC. They have been altered to specifically run on hardware that was made specifically for this, and even then management of it is not exactly simple. Not that I believe that 2003 Server will suddenly change that but just using Linux somewhere does not automatically make it the cheapest way.


The "standard" cluster these days is standard rack servers from a reputable vendor, along with a Linux distro tailor-made for cluster usage such as Rocks or OSCAR. Typically the only nonstandard hw, if any, is a high-speed network (Infiniband, Quadrics etc.).


And I believe the correct answer to your question is Traditionally it has been done by tuned versions of commercial Unices which added to the base cost of the OS over and above the very expensive custom built hardware.


Perhaps in the mid-1990'ies yes..


Recently Linux has become able to do many of these tasks by similarly being modified at a significant cost running on the same expensive custom hardware.


No. With the exception of the high-speed network card I mentioned above, the rest of the hw and sw are bog-standard. Of course there are exceptions, e.g. SGI, Cray, NEC, IBM etc. but then we're talking "real supercomputers" and not commodity clusters (the market MS is aiming at).


The recent HPC installation using mostly off the shelf parts (they didn't use Ethernet) was the one at Virginia Tech and that ran OS X, not Linux.


Not to piss on OSX, but Mac clusters are probably outnumbered 100:1 by Linux clusters.

Re:Too expensive my arse

[joib]

As I understand it this sort of thing can be done on just about any kind of computer. And at every university I've ever been to there's usually stacks of old pcs laying around.


As opposed to running email and word, HPC is one of these things where CPU power actually matters. Those 500 MHz PC:s aren't worth the hassle to set up and maintain. Not to mention that heterogeneous hardware (which a random bunch of discarded PC:s probably is) is a nightmare to maintain and program efficiently in parallel.

Most clusters consist of quality rack servers from a reputable vendor. TCO matters, not the cost of the hw alone.

Re:Too expensive my arse

[zcat_NZ]

When I first saw this on google news the headline was something about Windows for Supercomputers..

My first thought was "Oh, they've finally announced the real hardware requirements to run Vista"

Re:Too expensive my arse

[gentlemoose]

S'true. 1800 node renderfarms work just fine w/ gigE interconnects and NAS disk.

Re:Too expensive my arse

[BoneFlower]

Sure, it can be.

In addition to the issues with interconnects, raw performance of individual nodes, and heterogeneous clusters...

Reliability becomes a big deal with such old computers. Sure, a well designed cluster will be able to route around a significant number of failed nodes, but computing efficiency will plummet and won't be terribly predictable(often, predictability becomes more important than raw burst performance). You might have 20 nodes working today, 12 tommorow, back up to 20 for a few weeks, then lose 6, 2 of which are completely dead leaving you with 18 after the repair... You see the problem here?

That said, such things can make for interesting projects, and might make decent production systems in some contexts. They are not, however, a panacea to a universities HPC needs.

Speaking of learnign experiment HPC systems, does anyone know if any of the virtual machine solutions available for PCs can be used to create a software simulation of an HPC cluster? I realize such a thing would be near useless for real work, but might make for an interesting learning exercise and useful testbed before you deploy to the live cluster.

Re:Too expensive my arse

[Anonymous Coward]

OT (hence posting AC), but I love your sig.

Re:Too expensive my arse

[Anonymous Coward]

Dude, you're getting a super computer!

Re:Too expensive my arse

[Bert64]

And last i heard, microsoft's effort requires a 64bit cpu, so you have two choices:
Buy a whole heap of new 64bit systems, and then buy microsoft licenses for them.
Use a whole heap of old 32bit systems, and use a free install of linux on all of them.

Doing it microsoft's way requires significant up-front investment, while linux clusters can be constructed out of a collection of old/spare machines at no cost...
I build a cluster at work a few months ago, it was built using recently-obsoleted desktop systems that were otherwise sitting idle in a store room

Re:Too expensive my arse

[kestasjk]

You would usually call something that just assigns each computer it's own job and leaves them to it a grid, not a cluster. Cluster implies that it's a collection of computers acting like a single one.

Re:Too expensive my arse

[Weh]

Most researchers have limited funding

University researchers may have limited funding but a lot of researchers at large corps (oil/med/etc) don't have much trouble getting funds for their research, bear in mind also that not everything is research, for instance engineers may simply want to run some large numerical models etc. I have personally seen parallel processing on windows clusters implemented at a large corp, plenty of funding there.

Re:Too expensive my arse

[dbIII]

Real HPC needs highspeed, low latency internal interconnects and these are expensive.

It really depends on what you are doing with it. If the jobs can run in parallel well then gigabit networking is plenty. I'm looking after a couple of clusters used to interpret seismic data, and in some cases each node only reports back when you want it to do a daily checkpoint of where it is up to in a job.

Microsoft haven't even noticed clusters until now, so it will be a few years before anything of note is written to run on clusters on that platform - plus the whole poorly documented moving target operating systems which makes it pointless to port some software that has developed over decades and is run on clusters today.

Re:Too expensive my arse

[kesuki]

Some companies are in lockstep with microsoft, so this product will be beneficially to maintaining an 'all microsoft' deployment. yeah, having the slow interconnect speed limits the kind of clustering problems that can be done, but that's where all the linux clustering started :)

If you're working at a microsoft shop, and need clustering, this would be a product worth looking at, i don't know how many people that applies to, but i'm sure microsoft must have had customers asking for this, or else they wouldn't have all those stories about it.

Re:Too expensive my arse

[Anonymous Coward]

You are confusing general purpose clusters with HPC's. HPC's often require high speed interconnets and share data and processing time across all nodes to act as one machine. most general purpose clusters are not truly acting as one machine, they spread the load via connections or users or differing tasks rather than each node processing part of the job.

Re:Too expensive my arse

[multimediavt]

Ummm...I know of several clusters on our campus (VT) that are made of white boxes running Fedora, Gentoo, or Suse. One is a 200 node (400 CPU) Opteron cluster with a Myrinet interconnect named Anantham, and built by Dr. Varadarajan's graduate students. There are other smaller clusters ( 16 - 32 nodes ) of various design that are running on GigE. All of them built of white boxes and other off-the-shelf components ordered from mail-order companies. In the case of Anantham, all parts were ordered separately, i.e., RAM, motherboards, processors, cases, etc., and the system integration was done onsite. So, when you say, "Not to [sic] many are using Fedora or Slackware on some white box with parts from Best Buy to do HPC," I'm guessing you are referring to those in the TOP500 List? If so, yes, there aren't many that submit to the TOP500 List (from large sites) that are using a non-commercial version of Linux, i.e., RHEL. Many of the larger sites are going through first tier vendors (Dell, HP, IBM) for a turnkey cluster solution, but they are paying a premium for those systems for the sake of time to production. They could just as well buy white boxes, but they would be spending a great deal of their own time weeding out problem nodes and components that could be better spent on doing science and supporting users. Academia can afford to take the time to do this, DOE labs cannot, although, that paradigm is quickly shifting as academic budgets tighten and competition in the Computational Science and Engineering arenas heats up among research institutions.

Clusters (the topic of this original post) are not "traditionally [...] done by tuned versions of commercial Unices [sic]". Clusters are traditionally built with off-the-shelf components with Linux and specialized APIs and drivers for the interconnect being used. If you want to talk about HPC BEFORE 1998, then you are looking at large monolithic systems of a custom built nature.

System X does have a GigE network, but it is primarily used for management and job startup within the cluster. We have had a few users with specific MPICH2 needs that have used the GigE network for message passing, but the GigE network was not designed for that task. Our primary communication fabric is IB. We are currently running Mac OS X (10.3.9) on the system, but are evaluating alternatives.

Re:Too expensive my arse

[ortholattice]

Uh, sorry to be pedantic, but I think by "the electron flow around a trace" you mean the electromagnetic field around the trace. Electrons themselves never leave a printed circuit trace, i.e. the copper cladding "wire," but instead travel inside of it (or on its surface at high frequencies, resulting in "skin effect"). It sounds like they were modeling the transmission-line and crosstalk characteristics of the PC board.

Re:Too expensive my arse

[Antique Geekmeister]

There are exceptions: some folks do wind up digging up racks of old servers, at rock bottom prices or even for free, as their data centers or deployed installations decommission them. You can inherit quite a lot of slightly outdated hardware this way: if you can justify the electrical expense of running them, they're quite convenient for massive, lengthy computing jobs.

A lot of cluster managers also mistake "really expensive, physically robust servers for "will always be working". The complexities of such setups and the general frequency of failure of "high availability" software itself means that the much vaunted 99.99% uptime of such systems is usually based on serious cooking of the numbers, not any metric actually used in the field. After the crops of failures of things like the old IBM deskstar drives, the run of bad tantalum capacitors in Dell motherboards, and other failures that strike entire classes of brand new hardware, it's often better to use older, cheaper, burned in hardware that's had the BIOS updates and the kinks worked out, and save the extra money for the next round of upgrades in six months or a year.

Re:Too expensive my arse

[Antique Geekmeister]

Or upgrade something else you need a lot of for another project, and use the old 32-bit or 64-bit hardware with Linux as a free software testbed before pursuing a big hardware purchase. That gives you some experience with the tools before spending huge amounts of software and hardware money up front. And you can upgrade when you need to, since you've already got the cooling and power in place for the old hardware.

Re:Too expensive my arse

[fitten]

Well... running a single application that has multiple parts executed in parallel can be done on a cluster. Many embarassingly parallel applications are easily made into a manager/workers model and run on clusters (or grids). Because it is running one application (with parts of it running in parallel), it is "a collection of computers acting like a single one".

Re:Too expensive my arse

[ivan256]

A lot of times it's more a matter of latency than data volume... But still, you could be surprised by how low the latency can be over raw 100mbit ethernet, or using VI over ethernet. Fancy interconects are typically only for people who need high data rates *and* low latency, or for people who didn't do their homework.

Re:Too expensive my arse

[Anonymous Coward]

Yeah, I've found that vi works amazingly snappily over ethernet. Emacs on the other hand ...

*ducks*

Re:Too expensive my arse

[corvair2k1]

That's a pretty simplistic definition of a grid. A grid is typically a pool of resources that is not under a single administrative domain, which is transparently accessible as a utility. Resources on the grid can be clusters, file systems, single machines, etc. I think you're thinking that this would be a distributed application, where everything is under a single administrative domain.

A cluster is going to be managed as a single machine, true. But you're not necessarily even requiring communication to occur at all for different processes of a job on a cluster. You're basically saying that if you put embarrassingly parallel jobs on the cluster, it's not a cluster anymore.

The term cluster is seen more as an administrative characteristic than on what actual hardware is going into it. As long as you've got everything in one administrative domain, the computers are able to communicate with each other, and these computers are solely dedicated to doing jobs as given by the job scheduler, it's a cluster. Bonus points if there's a fast interconnect. However, there's nothing stopping anyone from submitting N single-process jobs to the N-node cluster, or an embarrassingly parallel job that only communicates at the beginning and end.

Re:Too expensive my arse

[Anonymous Coward]

I'd disagree to a certain extent. A cluster typically means a group of machines sharing a common architecture geographically very closely located (e.g. the same machine room). A grid is a concept that typically covers machines that are more widely spread, maybe even widely geographically distributed, and heterogeneous. In both cases, however, jobs on a grid or cluster may have significant interprocess communication or none, or alternatively be in a pipeline or workflow. In reality if there is a high degree of inter-process communication you want low-latency, high speed interconnects, which pushes you towards a cluster as this is where this type of communication speed and latency is available. For example MPICH is one of the familiar message passing communication libraries used for parallel programming on clusters, but there does exist an MPICH-G2 grid enabled version to allow you to achieve the same communication types (if not at the same efficiency) on a loosely coupled systems.

The word grid means different things to different people, but the key concept in most definitions is the idea that it is distributed computing in which the computing or data elements need not be geographically closely located. The idea of using the word grid is derived from an analogy to the electricity grid in terms the networked nature (to some extent) and a proposed ease-of-use equivalent to plugging in an appliance. One of the suggestions for a grid has been that the computing power should be available to an end user as if it was one - which matches your definition of a cluster.

In the end there is almost a continuum. If you have a series of Linux boxes racked up in a data centre connected via myrinet and operating via a job scheduling system like PBS then you probably have a cluster. If you have computers in the USA and Europe and are using Globus, then you probably have a Grid. If you have a collection of desktop machines, all of the same spec, operating together with Condor as the scheduler, then you have something on a continuum between the two.

Re:Too expensive my arse

[martinultima]

I am sure a little command line use doesn't stump them

Hell, I think that if anything, a command line would make things considerably faster – probably redundant, I know, but the guys running these would probably be a lot faster with a console editor like emacs or vi than with the likes of (insert favorite X- or Windows-based editor here), plus the system itself would be faster without the overhead of a graphical interface. And besides, supercomputers and clusters are meant to do computation, not fancy display output.

Re:Too expensive my arse

[Anonymous Coward]

Rarely these days do people make beowulfs out of desktop machines in this way.

To have a high mean time between failure you want to have a well-engineered machine. You also want to have the maximum
computing power per power supply to cut down on overall heat output and electricity costs. This means going for a dual
processor, dual core machine at a minimum these days. So that means that if you went to get a desktop machine of that
specification you are probably talking a minimum of $1000 or more. Desktop machines are bulky. Lay them on their sides
and they are about 4u. Putting them on shelves and networking them that way means that you have a practical space usage
of more like 8u. This space soon adds up. An alternative is to rack mount them, getting back to the 4u, but by this point
your individual unit cost is up to $1500. You can go to a second-tier supplier and get a 1u rackmount quad core machine
for around $2000 with a good support contract. This is excluding the per-unit cost of the high speed interconnects, racks,
gigabit switches, and the like.

Plus from a visual point of view a nice set of rack mounted 1u machines looks good.

The downside is, perhaps, that those 1u machines tend to be defeaning when you have a 6 foot rack full of the them.

Re:Too expensive my arse

[Anonymous Coward]

There are areas in which it is of use, even if beowulf clusters with easy-to-use software (OSCAR, Rocks, others) is available. This is where the need is to run Windows software because the existing software is built in a way that is windows-specific and a rewrite is not possible. Codes include (and this is in reference to the comments below about a beowulf being run at Motorola) can include chip simulation programs. This having been said, in most instances this would be serial code, in which case it is possible to run it on top of Linux using virtualisation software. In fact using virtualisation allows you the option of supporting various versions of Windows or customised virtual machines.

The main advatange of Microsoft's new offering is going to be explicitly parallel applications that require Windows. This is likely to be from taking existing Windows-only serial codes and parallelising them. How big a market this is I wouldn't like to guess, but I would hazard that Microsoft has done the research and feels that it is worth it.

This having been said it is possible to create clusters of Windows-based machines (e.g. using Condor or Sun Grid Engine to name two) and run MPI jobs and so on on them already. However an easy turnkey solution is going to be useful for big organisations
that don't the surety of something that will work with a guarantee from Microsoft rather than it being something that is
seen by the management as slightly more home-brew. I wouldn't like to try to add virtualisation into that mix too without a bit
of extra thought!

Re:Too expensive my arse

[x-caiver]

I'd have given you a +1 if I hadn't already posted in this thread this morning.

Re:Too expensive my arse

[jbplou]

Power is also another issue involved, that is not cheap.

Re:Too expensive my arse

[Anonymous Coward]

"But I fail to see how paying a "Windows" tax will make matters cheaper, or easier." - by geoff lane (93738) on Sunday June 11, @02:45AM (#15511887)

Simple: The best development toolset there is in Visual Studio 2005 (and, subsequent future versions, because it already is capable of this type of coding). It allows you to do things in it that are tougher to do in other languages, with greater ease via better abstraction.

That's how.

Re:Too expensive my arse

[mike_the_kid]

If you are really interested in getting the four nines you cite (or the 5 nines, aka sigma-6), the important thing is not that any one machine is running the whole time (that is, aiming for five nines on any one node is pointless.)

The trick is to try to have your system up and running optimally 99.999% of the time. This means recognizing that nodes will fail, building in redundancies, having a plan in place for maintenance, and having a plan to make sure the maintenance plan is followed.

You are correct in saying that this is not something you can buy just by picking the right brand. That said, I don't think increasing quantity is always the best path to quality.

Re:Too expensive my arse

[chthon]

I think User Mode Linux has been used for such setups.

Re:Too expensive my arse

[kl76]

Like bad management, they ordered a bunch of desktop PC they couldn't use, and no one authorized their return. So they sat around in unopened boxes until his team decided to make a Beowulf cluster...

The exact opposite happened at an academic HPC centre I used to work in - we ordered a small pre-built cluster from a prominent specialist German vendor, complete with snazzy custom rack, presumably at great expense. By the time it arrived, the guy whose idea it was had left, so it languished in the corner of a room for several months due to lack of interest. Eventually it got installed in the corner of an office and someone ordered some monitor/keyboard/mouse extension cables to turn it into a bunch of "desktop" PCs for staff use...:-)

Re:Too expensive my arse

[GmAz]

Just because Linux can do it doesn't mean people want to use Linux.

Re:Too expensive my arse

[turgid]

This is not aimed at scientists and engineers.

It's aimed at PHBs and Bean-Counters who pervade large corporate and public institutions and hold the purse strings. These are the very same people who hold back technological, scientific and financial progress due to a serious lack of clue.

And for us mere mortals...

[Bananatree3]

"but until now it has been too expensive and too difficult for many people to use effectively"

and what about the site licence needed for this baby, huh? For us mere basement-cluster builders, there is a cheaper, open source alternative: The OSCAR Project ( Open Source Cluster Application Resources). Yes, it runs on Linux, but it is a nearly step-by-step system of setting up HPC-level clusters. It is being used on many 100+ CPU High Performance Clusters around the world, and it is free without those pesky site licences.

Re:And for us mere mortals...

[Demetril]

Considering I work at one of the companies listed in the article MS is working with, and all our HPC clustering is working with linux, and will be for the coming future, I think this is primarily just a PR attempt. I don't even think we're remotely considered running anything on windows for our tasks.

Re:And for us mere mortals...

[FunOne]

I feel sorry that you think a 100+ CPU system is a High Performane Cluster....

Re:And for us mere mortals...

[Melkman]

Even a 4 cpu cluster can be a high performance cluster. As opposed to a high availability cluster, a high performance cluster is optimized for maximum performance instead of maximum availability. The number of CPU's or performance delivered has nothing to do with it.

Re:And for us mere mortals...

[Anonymous Coward]

My guess is that they'll adapt something like Analysis Server or Report Server to run on it and call it HPC. Then all the pointy haired bosses can say "Our data warehouse is running on a super-cluster".

What the?

[killjoe]

"but until now it has been too expensive and too difficult for many people to use effectively,' "

So the MS solution is cheaper then linux and easier to use the Mac clusters? I don't think so.

Ms is the "me too!" guy from the usenet. Everytime anybody else comes up with something Ms comes in afterwards and says "Me too!".

Re:What the?

[TheNetAvenger]

Not that I disagree with you on this topic, but your post is almost word for word what the industry said of Microsoft when it entered the Server market competing against Novell.

Microsoft was considered to be the 'me to servers' of the time, yet as it turns out the MS servers 'did' offer features that the Novell servers of the time didn't and application servers progressed to the point that MS kicked Novell's butt.

The push for application servers also opened the door for *nixes to enter back into the mainstream 'server' environments, as Novell was a pretty closed Server technology and applications running on the Novell server were a joke.

So if history repeats, don't be surprisd if MS does have an ace up its sleeve and its approach to the clustered server model using that ace and companies do find real advantages when using the Microsoft concepts.

However, even if MS does have an Ace, it would be kind of nice to see the technology envelope challenged, and see this back progression into other OSes and *nixes.

I guess it is the same old story, never underestimate MS...

Re:What the?

[killjoe]

While MS has had success with it's servers it has not been able to achieve the same monopoly status in that market that is has on the desktop and the office software markets. Since then they have tried repetedly to attain dominance in other markets by dumping software for free, forcing downloads with windows, paying customers to use them etc. Despite all that they have not been able to get better then third place in anything. Xbox, SQL server, sharepoint, great plains (what ever it's called now), their CRM software, etc are doing at best third place in the market. Other products have been utter failures MSN, MS at work, webtv, money, works, and several other web sites which they have ditched all got shot down in flames after sucking up tons of money.

MS has cleary lost a lot of it's mojo. They are not the MS of old. Sure they can still sink billions into products which have no market share till kingdom come but that may not be enough anymore. Alas MS has that luxury due their monopoly level profits on office and windows and other companies don't have that luxury. That's definately a disadvantage for any other company in IT. Still though MS is so incompetent that they are unable get SQL server to any better position then third place. They have to pay people to use IIS. I mean what does that say about them? Is that the mark of a juggernaut? I don't think so. I think it's a stink of desparation.

Finally I don't think MS has in it's heart to build products anymore. All they talk about is advertising. They are clearly in the process of transforming their company to be able to deliver advertising to windows users. I don't think things like cluster server are high priority at MS. It's just a last minute entry into the marketplace to try and halt the growth of Linux there. It won't work and MS just doesn't care all that much about it anyway.

Re:What the?

[TheNetAvenger]

I might be inclined to agree, but if you pay attention to the R&D of Microsoft, both of what is publically published and mass amounts of money into stuff that is unseen, I wouldn't count them out on anything...

They still put more into R&D than any other company, even IBM at its height ($$ or %).

This also isn't money flipped to crazy projects or ideas, but to some of the best engineers and theorists in the world.

Everyone here should drop by the Public portion of the MS R&D site once in a while, some of the concepts showcased are seeds of things that could be crucial in society next month or even 10 years from now.

Re:What the?

[killjoe]

MS has been cutting their R&D budget to make it's numbers look better. It's a recent trend and it's not likely to reverse itself. MS profitability is suffering due to deep discounts it has to make to combat linux and due to the flattning of their stock price (MS makes a ton of money buying and selling it's own stock). At this point they can not allow their stock price to drop at all because it would set up a chain reaction spiraling their stock downward and sending their profits to hell.

In order to avoid that they are cutting R&D to make their numbers look good.

It's a good tactic. It's not like their R&D has produced anything in the last five years anyway. Every product they have introduced in the last 10 years has been either an aquisition or a me too product. The research dept at MS is a wasteland where smart and talented people go to retire. The purpose of that dept is not to innovate, it's to remove smart people from the market.

Re:What the?

[ettlz]

Everyone here should drop by the Public portion of the MS R&D site once in a while, some of the concepts showcased are seeds of things that could be crucial in society next month or even 10 years from now.

Yes, they have some clever people working for them, but this is Microsoft we're talking about. They're a business, not an academic institute, and that worries me. And to claim that current R-and-D could be "crucial" in ten years' time is somewhat debatable; Microsoft seems to "make" things "crucial" regardless of necessity by strongarming the marketplace.

Re:What the?

[Anonymous Coward]

now this is pure FUD, no I take that back, you are spouting out and out lies. MS didn't cut R&D funding, They increased it by over a billion dollars a year, this is why the stock market dumped on them, they hate any company spending more money even if it is to potentially increase future earnings. MS profitability increased by 13% over last year and revenue increased by around the same. There is plenty to hate Microsoft over and pick on them for, but please use truths rather than pulling lies out of your arse.

Re:What the?

[Anonymous Coward]

Me too!

Re:What the?

[fitten]

MS has cleary lost a lot of it's mojo. They are not the MS of old. Sure they can still sink billions into products which have no market share till kingdom come but that may not be enough anymore.

I disagree. I think Microsoft has changed somewhat but I think they are still the Microsoft of old (same goals, etc.) It's the rest of the world that has changed the most (and thus, perhaps, why Microsoft has lost a bit of their mojo, as you say). There are *many* more computer users today than there were in 1995 and media regularly runs tech oriented stories so that today, you don't just see the huge stories... you see even smaller interest stories. Users are getting more familiar with computers and in very many areas, almost everyone has access to a computer of some kind and the Internet. More exposure, more media presence, more everything about computers today than ever efore. If you ask around in the USA, Europe, many parts of Asia, etc., the odds are that any computer user has at least heard of Linux, for example, and Apple's Macs have been getting more press now than ever... hell, Apple opens a retail store in NYC and it makes the every-15-minute run on Headline News. Before, no one knew much and if you got a PC, you got Microsoft's DOS or Windows. No one knew any better or anything else. These days, almost everyone knowledge of computers is a lot more broad than it used to be... largely thanks to the Internet and the other, more traditional, media.

Re:What the?

[fitten]

Ms is the "me too!" guy from the usenet. Everytime anybody else comes up with something Ms comes in afterwards and says "Me too!".

Sounds like today's Sony and the PS3 ;)

Re:What the?

[TheNetAvenger]

No offense, but you don't have the facts about MS R&D budget correct, let alone know much about the people or the projects and what has come out of them or you wouldn't discount MS R&D so easily...

Re:What the?

[TheNetAvenger]

Yes, they have some clever people working for them, but this is Microsoft we're talking about. They're a business, not an academic institute, and that worries me. And to claim that current R-and-D could be "crucial" in ten years' time is somewhat debatable; Microsoft seems to "make" things "crucial" regardless of necessity by strongarming the marketplace.


To make a distinction between 'academic institute' and business you must not understand what does happen at 'academic institutes' if you think it is all for the good of the people. Most R&D at academic institutes become businesses as well.

We are also at a time when the US Goverment has stopped pouring money into research and 'science', so we had better hope that companies like Microsoft help take up a portion of this slack or the 'next' big things are screwed.

Nanotechnology is already suffering greatly from lack of government funding in the US.

Also just because R&D work comes from a business, does not mean it is only ever used for that business. MS's R&D tends to be more of an exception to the rule, they are not like Sun's R&D where it is only stuff used at Sun, period.

There have been many things from MS R&D that are published and open for free use, even if it does not carry a standard 'open source' license. There is also a lot of work ouside the standard software engineering models, which MS has no interest in using itself and instead is technology given away to the hardware and other science markets.

Re:What the?

[killjoe]

"No offense, but you don't have the facts about MS R&D budget correct,"

It is you who has it wrong. Go check their latest filings with the SEC. You can lie to the press, you can lie to your customers, but you can't lie to the SEC and the shareholders.

"et alone know much about the people or the projects and what has come out of them or you wouldn't discount MS R&D so easily..."

Name five things that came out of MS reasearch in the last five years. Five things that are now MS products and are innovative. I dare you.

Re:What the?

[ookaze]

Not that I disagree with you on this topic, but your post is almost word for word what the industry said of Microsoft when it entered the Server market competing against Novell

But Novell is nothing like Linux and FOSS.

Microsoft was considered to be the 'me to servers' of the time, yet as it turns out the MS servers 'did' offer features that the Novell servers of the time didn't and application servers progressed to the point that MS kicked Novell's butt

You sound like features are only what made MS kick Novell's but. That's just misleading.

So if history repeats, don't be surprisd if MS does have an ace up its sleeve and its approach to the clustered server model using that ace and companies do find real advantages when using the Microsoft concepts

That's not history repeating here, but sth completely different. The situation you described previously is how Linux gained the HPC market.
MS to take HPC share from Linux based OS is sth completely different.

I guess it is the same old story, never underestimate MS...

I guess it's not. Never underestimate MS, like in consoles, mobiles, media centers, release dates, ... ?
I think you'd be more right than wrong by underestimating them.

Re:What the?

[TheNetAvenger]

Name five things that came out of MS reasearch in the last five years. Five things that are now MS products and are innovative. I dare you

In Vista alone, or you want to cover things from all the products from the last 5 years?

Indexing Algorithms like in SQL Server 2005 and Vista?
Flash Memory techniques for HD & I/O Caching in Vista?
Language Usage beyond Grammar (the new blue lines) in Word 2007?
Speech Recognition - The new speech engine in Vista?
GPU Multi-Tasking Technologies - like the WDDM model in Vista for both GPU and GPU RAM Virtualization?

Is 5 enough to satisfy the proof of your lack of understand regarding MS's R&D or shall we just make it 300 next?

Just off the top of 'my limited' knowledge we could go on with things regarding XHTML papers, the whole XAML constructs, the 3D scripting concepts from WPF, Windows Media Technologies (like you will use on your HD-DVD or Blu-Ray player) to 100s of other things...

I truly suggest you spend some time going through the public portion of MS R&D, there is a lot of stuff there that is demonstration of theory and other core technologies that will be a part of computing for years to come. MS may suck at business, but they have never hired stupid people when it comes to hard line engineering and theory.

MS Compiler technology is a very good demonstration that they have people that 'get' computing at its core and are very capable of using these concepts in practice. Oh, I didn't even touch on 50 or 60 things from R&D that is a part of Visual Studio 2005 compiler technologies, do you 'need' someone to point them out for you like a child also?

Re:What the?

[killjoe]

None of those things you mentioned are particularly innovative. They all have been around in lots of products and open source projects. By innovation I don't mean putting out a MS flavored implementation of some other technology. I mean something that is genuinely innovative.

Saying that vista has voice recognition when Mac has had for two years or that SQL server has a new indexing algorithm doesn't wash. Sorry.

Re:What the?

[TheNetAvenger]

None of those things you mentioned are particularly innovative. They all have been around in lots of products and open source projects. By innovation I don't mean putting out a MS flavored implementation of some other technology. I mean something that is genuinely innovative.

Saying that vista has voice recognition when Mac has had for two years or that SQL server has a new indexing algorithm doesn't wash. Sorry.

Just picky aren't ya?

A new way of indexing is 'innovative', but somehow unless this is something an end user can see and touch I don't think you will accept it as an innovation.

Computing and innovation exists beyond, end user touchable features. Complex code and the mathematics behind them are 'innovations' from an intellecutal standpoint and are 'new' ways of doing things. I'm sorry you don't hold the same definition.

As for all the other things I mentioned 'already' being available, that is not true. Find even ONE product or paper that address what the WDDM model in Vista is doing? There is none outside of Microsoft, even ATI and NVidia have not dealt with the new technologies 'MICROSOFT' introduced.

Also the 'way' the speech technologies in Vista work are new. Oh and BTW MS had Speech recognition back in 1992, far before you could get it on a Mac. (Go look up MS Sound System)

I'm truly not going to hold your hand and teach you what innovation is, or give you another 500 examples. There are tons of things MS brought forth that NEVER existed before that you are using on your Mac or *nix box as well. Even highlight a word and change the font to bold. That was a 'Select and modify' concept Microsoft introduced in MS Word for the Mac back in the mid 80s. Since then all GUI OSes have copied the select and modify model as well for interacting with objects and text.

But hey, MS just copies stuff in your world right?

Re:What the?

[killjoe]

"Computing and innovation exists beyond, end user touchable features. Complex code and the mathematics behind them are 'innovations' from an intellecutal standpoint and are 'new' ways of doing things. I'm sorry you don't hold the same definition."

Innovation by definition is something new. Something out of the ordinary. Something other people haven't thought of. If all MS has gotten out of it's research team for the last five years is speech recognition, database indexing, and a re-done XUL then they have wasted their money.

By your definition there is more innovation in postgres when SQL server.

"But hey, MS just copies stuff in your world right?"

Yes. And you have failed to mention even one thing that they didn't. Bold? You came out with bold? LOL

Re:What the?

[TheNetAvenger]

I apologize, I assumed you would be smart enough to understand. Next time I see your name, I will just pass out the crayons and move on...

Re:What the?

[killjoe]

You do that shillboy.

Re:What the?

[NullProg]

As for all the other things I mentioned 'already' being available, that is not true. Find even ONE product or paper that address what the WDDM model in Vista is doing? There is none outside of Microsoft, even ATI and NVidia have not dealt with the new technologies 'MICROSOFT' introduced.
WDDM is special? Explain why please, it hasn't been covered in detail in the ACM or Dr Dobbs yet. Any links?


Also the 'way' the speech technologies in Vista work are new. Oh and BTW MS had Speech recognition back in 1992, far before you could get it on a Mac. (Go look up MS Sound System)
Um, Microsoft Licensed the technology from Dragon. OS/2 Warp 4 had it before windows. Please provide links for proof of Microsoft using this in 1992.

Enjoy,

Re:What the?

[TheNetAvenger]

Um, Microsoft Licensed the technology from Dragon. OS/2 Warp 4 had it before windows. Please provide links for proof of Microsoft using this in 1992.


As I already stated, go look up 'MS Sound System'.

It was a sound card technology to push the industry (like creative) to release CD quality level sound for PCs with the introduction of the multimedia driver support in Windows 3.1. (I actually owned one of these when released and would play solitaire with only my voice. It included the sound card, microphone, and software.)

It included 'command' based speech recognition. To get 'dictation' level recognition at this time required a separate sound processor (which IBM was one company that produced such a product).

The current 'Vista' version of MS Speech recognition technologies are the result of Microsoft R&D work over the past 10 years.

MS's first 'dictation' level 'released' speech recognition software was a part of OfficeXP. This has been revised in several new directions, some of them quite novel and came from Microsoft R&D.

I would suggest anyone that is truly interested in this stuff to check out the Speech technologies at microsoft.com or research.microsoft.com or even do a few searches on the internet. This is not hard information to fact check.

Yeah, right

[imemyself]

but until now it has been too expensive

That's a good one. I doubt that the expensiveness of HPC is something that Microsoft is going to solve. MS's marketing folks could have atleast thought of something that *sounds* like it might be true.

And in other news...

[nurhussein]

...the HPC community of scientists and engineers continue to not care.

The same folks who operate nuclear accelerators probably don't have that much of a problem operating computers that they need Clippy and pretty colours to help them out "in case they get confused".

MS copies Apple AGAIN

[sakusha]

So how long has XGrid been available on the Mac? Two or three years?

Re:MS copies Apple AGAIN

[rritterson]

So how long has ROCKS been available as a linux distro? 4 years? 5? Apple copies Linux AGAIN.

Re:MS copies Apple AGAIN

[jcr]

Apple copies Linux AGAIN.

"Zilla". Look it up.

-jcr

Re:MS copies Apple AGAIN

[m874t232]

So how long has XGrid been available on the Mac? Two or three years?

Good question. After all, we all know that, to a Mac zealot, something doesn't count "innovation" until Apple has copied it from Linux or Xerox.

Too expensive?

[onlysolution]

"...but until now it has been too expensive and too difficult for many people to use effectively..." According to their licensing model EVERY machine costs 469 dollars... Meaning a 20 machine cluster would have a 10,000 dollar overhead just on the OS alone. Not to mention the fact that you'd be compelled to buy it again as Longhorn Cluster Ed. in just a couple of years... It seems like a little work setting up a free OS cluster would be a vastly preferable option, is there really any need or reason for this (at this cost anyway)?

Re:Too expensive?

[grammar fascist]

It seems like a little work setting up a free OS cluster would be a vastly preferable option, is there really any need or reason for this (at this cost anyway)?

Because nobody ever got fired for going Microsoft? There are certain people that will have warm fuzzies now that Microsoft has offered a solution.

I can't see that the intersection between that group and the one doing computationally-intensive research will be very large, though.

Re:Too expensive?

[Anonymous Coward]

Not to mention the fact that you'd be compelled to buy it again as Longhorn Cluster Ed.

That's because you'll need a cluster to run it.

Making up terms?

[Geeselegs]

From the microsoft site:
When did x86 get replaced with x64, I'm sure the 86 is to do with the number of transistors on the chip. Way to go microsoft.

Re:Making up terms?

[kupan787]

x64, I believe, just refers to the x86 instruction set for 64 bit processors. Also known as EM64T, x86-64 or AMD64 depending on who you talk to.

And the 86 has nothing to do with transistors on a chip (the 8086 had 29,000 transistors). It was just the part number of the first in a line of processors (8086 then followed by the 80186, 80286, i386, i486, etc...).

Re:Making up terms?

[VertigoAce]

I would assume the idea is to use the simplest designation possible for the 64-bit processors that consumers are likely to see. Saying AMD64 will confuse some consumers who have an Intel box that runs compatible code. Saying x86-64 is likely to just get shortened back to x86. EM64T is just too many letters and numbers with no obvious meaning. So x64 is simple and gets across the 64-bit aspect (so MS uses the term IA64 to refer to Itanium support in its server releases, where you can expect your customer might understand the difference between the two 64-bit systems).

Re:Making up terms?

[Anonymous Coward]

No. x86-64 != x64

Looking at this purely using Intel terminology (AMD has comparable terminology, but I'm still searching for their documentation pages):

x86-64 is the EM64T (Extended Memory 64-bit Technology) stuff. This is an extension to the existing IA-32 architecture. These processors merely run in IA-32e mode (with the submodes of compatibility mode and 64-bit mode). They don't necessarily use 64 bit address buses. I think the latest Intel and AMD offerings are ~48 bits, although I'm open to correction.

x64 is the IA-64 architecture - Itanium. This is real 64 bit.

Re:Making up terms?

[x-caiver]

Sorry, but you are mistaken.

x86-64 == x64

x64 is -not- the same thing as IA-64.
ia64 = Itanium, or as you put it "real 64 bit".
x64 = The 64-bit extensions ot the 32-bit instruction set that AMD came up with. AMD likes to call it AMD64, Intel likes to call it the bind-numbing EM64T, some people like to call it x86-64.

If you don't believe me, go download a trial copy of Server 2003 Standard x64, try installing it on your itanium, and then report back on how that worked out for you.

2003=2006?

[John_Booty]

It takes some serious marketing balls (and/or or a lack of marketing brains) to release a product branded "2003" when we're already halfway through 2006.

I actually have to applaud the naming move; it accurately lets everybody know that this product is based on Windows Server 2003. It would have been quite misleading if they'd passed it off as " Windows Compute Cluster Server 2006".

Wonder what the meetings between the marketing team and the engineering team were like for this one. :)

Re:2003=2006?

[MadTinfoilHatter]

It would have been quite misleading if they'd passed it off as " Windows Compute Cluster Server 2006".

...or Windows ClusterFsck Edition. Oh wait, that's Vista. Nevermind.

Re:2003=2006?

[Lost Race]

The engineers have zero input, or maybe slightly less, into the naming of the product. I don't recall ever being in a meeting with marketing people.

Maybe I'm paying too much attention to detail...

[Ethan Allison]

Why is it called Windows Compute Cluster Server 2003 if it's ... um, 2006?

cluster=collective? a good match with the icon! XD

[plasmacutter]

resistance is futile, you shall be assimilated into the windows cluster ; )

Re:Maybe I'm paying too much attention to detail..

[Vrykoulakas]

See Above ^

Absurdity can be profitable"

[asky]

"...but until now it has been too expensive and too difficult for many people to use effectively."

I know many people take exception to that remark. But not everyone knows how to build Beowulf clusters.Some of us thought it was insane when, in the 90s, Microsoft said they were going to enter the server market. Yet here they are. And who in their right minds would run their web services out of IIS? (Then again, Apache now runs on Windows.)

The point is, just because the idea is absurd doesn't mean it won't happen. If corporate consolidations put support for technical computing under the IT department, and support for Linux is considered toodifficult for the IT folks, it's only a matter of time before the decree to port technical computing applications to Windows.

The fact is, M$ has access to software vendors, hardware vendors, and large customers in ways that Linux companies do not. They can create markets where they shouldn't be justified (unless you think all operating systems really require anti-virus software).

I'd love to be wrong about this. But I've finally come to the conclusion that sound technical judgement does not stop absurdity from happening.

Re:Absurdity can be profitable"

[nacturation]

And who in their right minds would run their web services out of IIS?

eBay for one.

Re:Absurdity can be profitable"

[misleb]

I'd love to be wrong about this. But I've finally come to the conclusion that sound technical judgement does not stop absurdity from happening.

Fortunately, I think the bottom line will prevent the absurd from happening in this case. If I understand correctly, licensing for each node will cost something like $465. That can really add up. And given existing free and relatively mature alternatives (Rocks, OSCAR), I don't see the draw of the MS solution. The main reason MS was able to penetrate the server market originally was because they owned the desktop. They have no such foot in the door here. Compute clusters are largely isolated systems, AFAIK.

-matthew

Re:Absurdity can be profitable"

[VividU]

And who in their right minds would run their web services out of IIS?

I do. I run Linux/Apache too. And despite the dire warnings of my Linux consultants who advised me not go near IIS, guess what happened? A whole lot of nothing. 100% uptime (so far) and a whole lot of crickets.

Re:Absurdity can be profitable"

[pwhysall]

Only sort of.

The front end web servers are IIS. The business logic is all Java, Solaris and Oracle.

Re:Absurdity can be profitable"

[Anonymous Coward]

A whole lot of nothing. 100% uptime (so far) and a whole lot of crickets.

Wait till you connect it to the net...

Re:Absurdity can be profitable"

[Anonymous Coward]

And who in their right minds would run their web services out of IIS?

Everyone who is not stuck in the past. IIS6 is actually very capable, reliable and have had very few security incidents (fewer than Apache2).

Re:Absurdity can be profitable"

[KwKSilver]

That explains why I haven't gotten the last item I "won" almost three weeks after paying up.

Re:Absurdity can be profitable"

[nacturation]

That explains why I haven't gotten the last item I "won" almost three weeks after paying up.

Of course. Had the servers been running Linux, the seller you bought from would naturally be more honest! :)

Another choice: Rocks Clusters

[joib]

Another, perhaps even more popular Linux cluster distro is Rocks Clusters.

While I don't have personal experience with OSCAR, Rocks is really good. These days, doing a cluster with a "normal" distro is insane. I think MS will have to think long and hard before they come up with something equally easy to install and manage as Rocks.

That being said, I think MS is not targeting Win CCS at academic supercomputing, which has a long history of using Unix/Linux, but rather they want to expand HPC to business customers who otherwise have a 100 % MS environment.

Re:Another choice: Rocks Clusters

[Anonymous Coward]

Probably the Microsoft salesfolks will get some sales out of stories like the above two...
"Look, even the small HPC market is fragmented on Linux. You will never know if what you are using is the best one or you should have gone to the other distribution. Nobody will be fired for choosing Microsoft!"

Re:Another choice: Rocks Clusters

[icedevil]

Why is it insane? Setting up a cluster with any distro is trivial. Granted OSCAR and Rocks are nice because they're geared towards clusters but that does not make something like Fedora or Suse difficult to use in a cluster configuration by any means.

Re:Another choice: Rocks Clusters

[ShakaUVM]

Nice to see the Rocks Clusters project is still going strong. A buddy of mine F. David Sacerdoti did a lot of work for it back in the day at UC San Diego / the San Diego Supercomputer Center.

Re:Another choice: Rocks Clusters

[joib]

Why is it insane?


Because you'll waste plenty of time solving problems that the Rocks team has already solved. See this presentation. Especially slides 38 and 47-57.

Re:Another choice: Rocks Clusters

[Anonymous Coward]

How's Sacerdoti doing these days? I didn't really know him myself, but heard his fiance (?) was heading towards Yale, and he was intending to head to the East Coast himself, looking for things out there. Sadly, Yale's HPC endeavors weren't terribly cohesive at the time. :|

Re:Another choice: Rocks Clusters

[Frumious Wombat]

I'll give the Rocks guys a plug (and not just because they were great guys when I was setting my first clusters up), for the elegance, and portability, of their solutions. The configuring nodes through XML scripts, scalable authentication system, support for I386/x86-64/IA-64 simultaneously, auto-configuring drivers for Myrinet, correctly configured Sun Grid Engine with no end-user interaction to set it up, and the funky solutions such as the Display Wall, makes them a stand-out distribution.

On the other hand, there are places, such as Cornell's Theory Center, which have 'successfully' gone to windows clusters. I personally think it's the Wrong Thing, given the difficulty of porting to Windows from Unix, and doubt it's as easy to set up as Rocks, but they're making it work. More likely, this will be used in Windows-heavy environments for high-throughput computing, rather than high-performance.

Re:Another choice: Rocks Clusters

[x2A]

"MS will have to think long and hard before they come up with something equally easy to install and manage as Rocks"

MS introduces New Windows 2003 Paper(tm)!

(just watch out for Mosix Scissors)

Re:Another choice: Rocks Clusters

[Anonymous Coward]

You've never tried this, have you? Setting up ROCKs is easy and managing it is not too bad. Trying to get all the stuff they do automatically right on your own would be an incredible waste of effort unless you have some very specific (and non-typical) requirements. ROCKs simply rocks. (It takes about 1 hour to setup a 40-node fully-functional cluster.)

Re:Another choice: Rocks Clusters

[Phroggy]

I've been pretty happy with Rocks so far. It's not perfect, but it looks well-designed, the documentation is friendly, and there's a mailing list.

The First???

[Anonymous Coward]

"The software is Microsoft's first to run parallel HPC applications"

This is just plain wrong.

Two years ago, I took a mini-course at Cornell Theory Center on how to set up a Windows 2000 computational cluster with MPI. MS has had clustering software for some time (ie, since Windows 2000).

Re:The First???

[Anonymous Coward]

I did manage to flummox the professor with the following question (I'm being serious here!):

Why does a 256 or 512-node computational cluster need a copy of Windows Media Player on each machine?

It was then that I realized that MS's approach to operating systems targeted to different applications is not to strip anything out, but to add layer upon layer of extra functionality to their basic "home computing" OS.

Yes, friends, their Windows 2000 computational cluster software does indeed ship with Media Player.

I can't speak of SQL Server or other variants: does anyone have a clue?

Re:The First???

[Bert64]

Does it also need a hard disk in every machine? Imagine all the extra heat produced by 500 superfluous hard drives. All the clusters i've built have been diskless, and booted from the network (or from a floppy, i did produce such a ghetto-cluster from old hardware once as an example)

Re:The First???

[agallagh42]

I don't know about booting from the network, but boot from SAN is fully supported. There's no need to have local disks in any Windows server.

Missing the point a bit?

[nixascyborg]

The idea behind Windows clusters being 'cheaper' has nothing to do with the individual price of the OS (versus, for instance, free Linux); the named price is low, not free, but that is not the point of your savings with a Windows HPC cluster. The point is that most programmers work on a Windows platform and have experience with it. And if you program with/for Windows and, for instance VS 2005, MS counts on the effort of building programs that run on HPC to be considerably less effort than it is on a Linux (or Xgrid) cluster. Making existing Windows 'hits' clusterable (i heard mention somewhere of image, movie and 3d processing software) is easier because of this too; making it work on other clusters is a pain because there you would have to work in an environment you are not used too. Like all things with MS; they count on the familiarity and ease of use to make this all run. That is what makes it cheaper; you cannot get a Linux HPC programmer and if you find him/her he will be godawful expensive; for WinHPC it will just be 'another VS programmer' of which there are a lot. Look for MS to add testing, debugging and development aids for HPC in the upcoming versions of VS.

Re:Missing the point a bit?

[nixascyborg]

What I forgot; ofcourse, getting linux admins to support the cluster is difficult also; Windows admins (although quite bad usually) are growing on trees; they can install and 'maintain' it. When stuff goes bad they usually cannot solve anything, but the company you're workign for will not think that far ahead. Setting up a WinHPC cluster is easy if you have the right hardware prepared; I guess the fact that it is 'just another Win2003' is reason for IT directors to call it 'cheaper' than Linux, considering the fact that maintanance is always an order of magnitude more costly than buying software.

Re:Missing the point a bit?

[waferhead]

"...IT directors to call it 'cheaper' than Linux, considering the fact that maintanance is always an order of magnitude more costly than buying software."

Tell me what's cheaper when your network gets overrun by a nasty virus, that infects all your machines, starting with the servers, and starts wiping drives, and end up costing unimaginable work and funds to even start to fix.

Please don't whine that "That's impossible!!!" or any such BS, it has HAPPENED, and is likely to happen again.

Monoculture sucks.
Thank MS for all your potential cost savings.
You pay for what you get perhaps.

(...And for some inexplicable reason, many if not most of the places that got hosed the worst are STILL 100% MS shops. I'm not bitter or anything...Concerned.)

Re:Missing the point a bit?

[Nightspirit]

If an IT person has gotten a windows virus since 2002 he should be fired. I have never gotten a XP virus, and 2003 server is supposed to be more locked down than XP, so you would have to be a royal moran to get a 2003 server virus.

Re:Missing the point a bit?

[nixascyborg]

How is this comment relevant for HPC ; you're not going to expose anything on the web directly and you are not going to browse (pr0n) sites on those machines or read mail etc? It will likely be an enclose development / OTAP environment which makes the chance for viruses very small.
Also (slightly offtopic, but still relevant) ; we run most servers on Linux, most desktops on Windows in our company and all of this since 2001 with > 150 people; we have not had ONE virus on either desktop or server. If you are a bit smart you have all incoming/outgoing virusscanned, on mail, smb and webdav. If your Win admin has half a brain, you can work with Windows fine without encountering any problems, ever.

Re:Missing the point a bit?

[Bert64]

Well that depends...
Will they force you to have IE and outlook express installed on every node without the ability to remove them? If there is a browser installed, then it will be used to browse at some point just because it's convenient to do it from the console rather than walking back to a proper workstation.

Also, does this support serial consoles? Imagine the hassle of having to hook up screens to every node for maintenence...

Re:Missing the point a bit?

[killjoe]

you have presented the MS argument masterfully. Are you a professional PR person? Do you work for a PR or ad agency? I suspect you do.

Anyway your points are not really that relevent because they are not that true. HPC applications tend to be very complex. They are not the types of applications you are likely to trust to a bunch of VS monkeys who draw GUIs with bound controls and call it an application. As for administration there is a mountain of evidence that Nix administration is cheaper then MS administration. Studies have shown that the average linux admin administers significantly more amchines then your average windows admin. They may cost more but they admin more servers. Studies have also shown that windows shops tend to have more servers per user then unix shops and that unix servers tend to run more applications per server.

Ask anybody who has done both and they will tell you. It's much easier to maintain a mass of unix servers then a mass of windows servers. Have ever tried to back up a bunch of windows server? Try doing it without spending ten thousand dollars.

Re:Missing the point a bit?

[nixascyborg]

:)
I am just saying what MS uses to sell this (and their other) product(s). If it is true or not is simply not relevant; their way of getting market works like this and it works well, because a lot of people are ignorant to what they buy. Even if the costs is $millions; we see it happening every day in a significantly large company.
I don't work in PR; I am a programmer working mostly on with Java, Ruby and PHP. My background is C and in university I did program a Unix HPC environment (based on Solaris at that time); we had to implement several algorithms in C to run on that cluster. It was difficult, that is exactly why I think the MS arguments will actually work with managers.

Re:Missing the point a bit?

[Vengeance]

Of course, when your HPC starts acting up in the middle of a massive simulation, 'reboot the cluster' isn't really a good answer, is it?

Re:Missing the point a bit?

[Tjp($)pjT]

for WinHPC it will just be just another VS programmer of which there are a lot

And those "another VS Programmer types will know all the pitfalls that are relevant to distributed computations in a cluster environment because of ... It is not the environment for development that is important. XCode for Mac is not really hard to learn. Heck, even at MS on campus, most large projects eschew the IDE in favor of command files that invoke nmake directly. (And I worked in Visual Studio Enterprise Edition, so have first hand knowledge, as well as having done a stint earlier with Visual Basic.) This is all marketing hype to make it appear that MS is on top of the game, when it has been quite some time since they created a first mover product. They usually sit back and watch, and then either build a product in an existing market that shows promise or buy a company to acquire a product in the market space of interest. Nothing new to see here... move along ...

Re:Missing the point a bit?

[nietsch]

So actually you are not one of those plentyfull VS programmers that you can pick up by the dozen at the local community college? ANd you have never worked professionally with this super-easy VS that makes programming so easy that even stupid managers can understand it?
Somehow I don't think it is a suprise that you blindly copy MS-FUD.
There is no magic bullet when it comes to programming. You cannot develop smart programs with dumb people. You may wish you could, but that is probably because you are one of them.

Re:Missing the point a bit?

[killjoe]

You have never worked with VS and yet you think it will be easier? Why?

Managers are not influenced by rationality, reason or facts. They are influenced by golf games, junkets, and gold watches. That's how MS wins business. See godaddy for example. MS paid godaddy to switch their parking to IIS.

Money talks and the CIOs love gold watches and golf junkets in Hawaii.

Re:Missing the point a bit?

[MP3Chuck]

"you're not going to expose anything on the web directly"

Not going to and not supposed to are two entirely different things. ;)

Re:Missing the point a bit?

[x2A]

"You have never worked with VS and yet you think it will be easier?"

Try reading a post before you reply to it. It begins with "The idea behind..." and discusses arguments that MS uses to sell, and that the arguments often work and MS gets the sale. At no point does he make the opinions out to be his own, or give any real indication or whether he agrees or not. Just "What MS says == What PHB's like to hear", which is actually true a whole lot of the time.

Re:Missing the point a bit?

[x2A]

"Somehow I don't think it is a suprise that you blindly copy MS-FUD"

Well it would be pretty damn stupid to talk about what MS says to sell their software, without copying what they say, wouldn't it? "The idea behind Windows clusters is that they use giant rats that have been genetically engineered to perform complex calculations" would be ya know... a lie. That's not the idea behind it at all. The idea behind it is that you CAN save money in other areas greater than the cost of the actual OS. Sure, maybe that doesn't actually work out to be true, no one here is saying that it does, but that's the IDEA behind it. The idea behind something doesn't actually have to work out true for it to be an idea behind something. D'ya get yet???

Somehow I don't think it is a surprise that you blindly copy /.-FUD.

Re:Missing the point a bit?

[moria]

There is nothing called "Linux HPC Programmer" or "Windows HPC Programmer". HPC has been, is, and will be the area of progammers of C, FORTRAN and some other special-purpose languages. There is no need for GUI, or .NET, or anything else M$ offers for general-purpose programming. Actually, there are so many things only available on UNIX and X11, like special-purpose visualization programs built on X11. If there are some "Windows HPC Programmers", there are much less of them than there are "Linux HPC Programmer", and since it requires so much expertise to port so many special-purpose UNIX-only tools, I bet they will be much much more expensive, too.

Programming tool is not something that can make a difference in this case. Having worked on supporting non-computer engineering people to write HPC programs, I know a lot mechanical engineering people writing programs on Windows, upload it to the cluster, sometimes via nice web interface, and get them compiled to automation tools.

There are 2 types of HPC programmers: computer experts and engineering/science people. For the computer experts who mainly support the engineering/science people in writing parallel programs and who write tools to simplify the use of HPC, most of them are all real geeks and know the UNIX way of doing things is much much more effecient; and for the engineering/science people, you cannot say letting them to learn VS 200x is something easy.

Sorry, MSFT, but HPC is the field having a high concentration of UNIX people and is the field that you way of cheap programming might not work.

Re:Missing the point a bit?

[Anonymous Coward]

The point is that most programmers work on a Windows platform and have experience with it.

This sounds really wrong - I'd like to see a qualification of that statement....

In terms of windowing environments, I'd guess that a lot of poeple are skilled in using MFC etc... But HPC isn't exactly about GUI apps - I'd expect that a lot of the backend code is just plain C, C++ or maybe a higher level language.

Re:Missing the point a bit?

[nietsch]

Usually people post their own opinions here, not copied from an MS ad. Sure you can use some of MS' pov, but if you add nothing of your own you are just repeating PR crap, AKA FUD.

Re:Missing the point a bit?

[BigFootApe]

The point is that most programmers work on a Windows platform and have experience with it.

The intersection between the skill-sets of Win32 app programming and HPC programming is approximately 0.

Re:Missing the point a bit?

[x2A]

Very true. Shame on him for posting the other side of the argument, to slashdot of all places, what an idiot.

Re:Missing the point a bit?

[FellowConspirator]

Most scientific/engineering programmers (myself included, though technically I'm a computational scientist) do NOT work on a Windows platform. Windows APIs could not be more foreign. Most users of HPC computing are purely UNIX (and more recently Linux) developers. Windows has never been the platform of choice for high-performance technical computing, and frankly, it never will be -- it's not designed for it (even Windows HPC isn't, marketing notwithstanding), applications are not developed for it, and MS doesn't even pretend to market to segments that make the most use out of it (never appeared at BIO, PSC, ISMB, etc.).

Screen Door on Submarine 1.0

[carcosa30]

Like this will ever fly.

It's rare that I've heard anything as ridiculous as this. There's nowhere UNIX is more entrenched than in big iron computing. Not only that, but the users-- by definition-- aren't dumb. Microsoft don't have a snowball's chance in hell in that marketing space, and I expect this is going to make Microsoft Bob look like a hot seller in comparison.

Re:Screen Door on Submarine 1.0

[Anonymous Coward]

You should get out more often into the real world.

The Cornell Theory Center, the leading academic resource for MS Windows cluster computing, holds classes at two locations:
(1) the Cornell campus in Ithaca, NY
(2) their off-campus facility on Wall Street.

Cluster computing is used heavily in stock, financial derivative, currency trading etc. MS couldn't care less about Linux geeks doing scientific supercomputing. They're going where the $$$ is.

A winner is you.

[RoffleTheWaffle]

Finally, an alternative operating system for that darned Playstation 3 personal computer. I thought I'd never see the day.

Seriously though. Anyone who buys into this needs to go sit in their failbox and think about what they've done. This kind of stuff has been done on the cheap with Unix and Posix operating systems for ages. I guess this could come in handy for experiments observing how artificially bloated clusters function, though.

Re:A winner is you.

[Antique Geekmeister]

I've strongly suggested Xbox based Linux clusters: besides the fun of making Microsoft lose money on every console purchased, then not buying any games for 200 machines. Unfortunately, the heat and poor venting of those power supplies is a problem: I wouldn't want to run more than a few in a small space, but it could also make for a fun cluster computing demo. Dual-boot one of them for games to draw people to your booth at a trade show, and you can have a bit of fun irritating Microsoft while doing actual cluster computing.

and what is new about?????

[proudhawk]

uh gee. I think linux, BSD, solaris and others
had HPC capability years ago.

it took Microsoft long enough, huh?

Re:and what is new about?????

[x2A]

"it took Microsoft long enough, huh?"

It took them this long to get this release out yes, but no one said this is their first cluster software release though, think you might have imagined that bit because it's what you want to read.

And on the 2nd Tuesday of the Month...

[Anonymous Coward]

... your cluster grinds to a halt as it runs Windows Update!

No thanks Bill.

Re:Absurdity can be profitable

[mikeb]

"But I've finally come to the conclusion that sound technical judgement does not stop absurdity from happening"

Something that the majority of Slashdot readers seem not to understand (and with justification) is that purchasing decisions are not rational.

A basic training course on sales techniques will, unless it's totally bogus, emphasise the fact that purchasing is based on emotion, not rationality. Some 80-90% of all sales are emotion-driven and then sometimes post-facto justified by selectively picking facts.

As the world becomes a more complex place and huge amounts of information become available to prospective purchasers there's a kind of paradox emerging that will horrify economists who cling to the theory that perfect markets are based on rational purchasers with perfect information, because the reverse is happening.

Most purchasers are not analytic personalities. People who hang around Slashdot underestimate how much they have (in general) honed their own analytic skills with years of practice while most middle-tier managers in corporates never did. For those non-analytic people, being asked to rationally evaluate a mass of facts and statistics is a SCARY proposition. That's not how they got their job, they did that by looking good in a suit and licking backsides more or less assiduously whilst being ok at judging how the politics are shaping up. Their skillset is way different from yours and they react differently.

The more information you make available to those people, the less they are likely to use and the more they will look around for 'safe' decisions. This will be especially true if their promotion prospects may depend on the outcome. THEY ARE NOT SPENDING THEIR OWN MONEY, it's the company's. Their decision will be based on the likelihood of retaining their job or getting promoted before their mistakes are discovered.

So, figure for yourself. On the one hand some technical guy they distrust because he's smart can 'download an ISO from the interweb and build a cluster myself' or 'buy from Microsoft'.

The first bit of irrational figuring will be 'the Microsoft stuff costs tens of thousands but the geek says it's free - that does not compute, he must be wrong'. The second will be 'if it goes wrong who will get the blame'. Guess the outcome of that one for yourself.

The result is fairly predictable IF you understand the parameters. Microsoft's marketing does understand where it's operating and will be well aware that its customer base is heavily loaded with irrational people. Most likely they are hearing squeals from that customer base asking where Microsoft's compute cluster solution is because 'we want to buy one'. It would be foolish not to give them one surely?

compliance to standards?

[WinEveryGame]

Per Microsoft document:

"Microsoft Message Passing Interface (MS-MPI) implementation is fully compatible with the reference MPICH2"

I guess given the fact that Microsoft is pathetically behind Linux when it comes to high performance computing, they may actually play by the rules here.

Anyone has an insight on this one? Do they have a API lock-in strategy here as well?

Re:compliance to standards?

[Anonymous Coward]

Magic Monopoly Ball says: Check again in five years.

Re:compliance to standards?

[Anonymous Coward]

Do they have a API lock-in strategy here as well?

Of course! From the FAQ:

Q. How does a Windows-based compute cluster compare with a cluster running UNIX or Linux?
A. There is little substantive difference, but UNIX-based solutions should be fully ported to Windows to realize the full benefits of the Windows operating system. There are several differences between UNIX-based operating systems and Windows. For example, I/O operations and threading are different on UNIX-based systems than they are on Windows. I/O intensive applications will benefit from using Windows native I/O APIs rather than UNIX style I/O APIs.

Re:compliance to standards?

[Bill Barth]

Actually they have an API lock-out strategy:

MS-MPI is (last I heard, at least) incompatible with Windows Services for UNIX, so if your MPI code is POSIXy, you're going to have to port it over to Win32 before it will run on this new product. You don't have to rewrite the communications bits, but your I/O will have to be rewritten.

Re:compliance to standards?

[x-caiver]

Not quite right. Actually Services for UNIX is incompatible with the x64 versions of Windows Server 2003, which is the required operating system family for the Compute Cluster.

Server 2003 R2 (http://www.microsoft.com/windowsserver2003/r2/wha tsnewinr2.mspx) adds the "Subsystem for Unix-based Applications" (SUA: http://www.microsoft.com/windowsserver2003/r2/unix interop/default.mspx) . You can either install R2 on one machine and recompile your app to make a win32 version, or you can install R2 on all your compute nodes to get the SUA environment everywhere. (the first option would probably be better)

Re:compliance to standards?

[x-caiver]

Yes Microsoft is playing by the rules. In this case it is best for both the community (not much work to move over) and for Microsoft (not much work for people to move over).

The MPICH2 code is the reference implementation being used, and that's good for everybody. Another tidbit that is good for everyone - Microsoft is even helping non-Microsoft users of MPICH2. As amusing as it is to say, Microsoft does have some expertise in Windows programming and has learned a few things about writing solid and secure code (yes yes, ha ha, but seriously there are some sharp people at the company regardless of what people think about Clippy and Longhorn). Fixes that were made to MPICH2 are being shared back with Argonne National Labs.

I don't think anyone here is going to argue that the people at Argonne are stupid, so if they think they changes that Microsoft's high performance computing team made are good, then I'd say that's a pretty good thumbs-up to the MS team responsible.

Re:compliance to standards?

[Anonymous Coward]

Microsoft hired lots of the people who originally made this stuff happen presumably in the hope that like German WW2 rocket scientists their brains could be picked for good ideas in the future. In the case of my ex-boss's old boss, he's still getting the Einstein treatment in Redmond. He doesn't produce anything valuable, but he was expensive to buy so no-one's considering letting him go. After a few more years he'll retire on a fat bonus and the only loss is that he didn't get to spend a few more years telling horror stories about wire-wrapping and hand assembled machine code to teenage hot-shots who think a Vax is a type of cleaning equipment.

Or maybe he does, Microsoft has plenty of hot-shots even if most of them aren't teenagers any more.

Re:compliance to standards?

[Bill Barth]

It's my understanding that R2 and Compute Cluster Edition are incompatible. Has that changed? Also, is it possible to make a statically-linked binary under Windows that uses SUA? If not, you'd have to have it everywhere.

Re:compliance to standards?

[x-caiver]

You're right that there is no Compute Cluster Edition R2. I should have been more clear and said that you'd need to get either the x64 Standard Edition R2 or x64 Enterprise Edition R2 to get the SUA feature.

As for static linking... I think that's how you are supposed to be able to do it (in which case you'd need only a single R2 install, rather then ever node being R2), but I've never tried myself.

Not for everyone... but think ISVs

[WoTG]

Ok, this isn't exactly going to be a hit with the scientific HPC community who already have all the clustering software that they need. But, think about MS's best customers, corporations. Imagine a new scheduling module for an ERP. If the model is complex enough, and if it has enough components and rules, it can easily become a major burden for a single server. And no, database clustering isn't necessarily the same -- not everything can be coded as a SQL statement, and even if they can, it isn't necessarily a smart way to apply a particular algo to a set of data. A Microsoft Windows based HPC unit would be perfect for the independent software vendor to use to power their new module -- assuming of course that the ERP itself runs on Windows. Odds are good that at least the client-side application is Windows compatible.

GNU/Linux is better for clusters

[J4nus_slashdotter]

If you use GNU/Linux for your cluster, you won't pay any licence. So, for the same cost, you can buy more computers and you will have more calculation power, so your cluster will be efficience..

Re:GNU/Linux is better for clusters

[hackstraw]

If you use GNU/Linux for your cluster, you won't pay any licence.

AND, you get the source. HPC guys are known to tweak the TCP/IP stack and/or drivers. Can't do that even after paying the bargain of ~$500/node for their OS.

Re:GNU/Linux is better for clusters

[dlapine]

Yes, we do :).

We tweak the kernel for long-haul IP traffic, certain types of error messages (thousands of warnings per thousands nodes equals gigabytes of logs per day, otherwise), and other things. We also update and build custom drivers for our gige and fiber channel cards. All of which isn't possible with the Windows versions.

We also have support contracts from our software vendors (IBM, Redhat and SuSE) in case we run into issues where some other professional opinions and expertise make a difference. We tend to get quick answers and good solutions from these vendors- I have no idea what dealing with MS tech support would be like.

Heh!

[jd]

Microsoft delivered the keynote speech for the Supercomputer conference SC|05 in Seattle last year and had a huge stand in the main hallway. Most people were walking straight past it and were gathering round the stands offering free Linux CDs. (I would really like to see at least one of the BSDs to add something like bproc - Linux has some amazing capabilities and those can and are used in many applications, but NetBSD has better I/O throughput and there are many cases where cluster applications are I/O-bound rather than CPU-bound or feature-bound.)

Microsoft's MPI implementation is, if I understand their materials correctly, based on MPICH (a BSD-licensed Open Source product) with some in-house fine tuning. MPICH is a good reference implementation but is not terribly fast and is getting to be long in the tooth. Far as I know, it doesn't have much in the way of fault tolerance in it, either. LAMPI and OpenMPI are built for speed (although I've found OpenMPI has room for substantial improvement) and have some fault tolerance support. So, they don't seem to be using an amazing architecture.

Last, but by no means least, Microsoft's freebies were limited to an Opteron-specific Windows 2003 Cluster Edition beta and a cookie. By comparison, many others had booklets on what their products did, papers on the theoretical work being done, working demos (the molecular modeler with forced feedback was amazing) and some highly knowledgeable geeks to answer detailed technical questions.

Microsoft may - someday - be an interesting player in the cluster market. Right now, though, they really don't seem to get what it is all about. I'm not trying to bash Microsoft here, they really don't have a product that is useful for the high-performance market, and seem to have the wrong libraries and interfaces for using the servers in a load-balancing, fail-over or distributed storage environment. This isn't to say the other vendors were perfect - I saw many areas that were horribly inefficient and poorly implemented - but rather that Microsoft would have done better to have come back from the show and re-thought what it was that they wanted the Cluster Edition to do.

Re:Heh!

[Anonymous Coward]

Microsoft may - someday - be an interesting player in the cluster market
You think so ?? I don't, they couldn't even deliver a working cluster solution on a Digital Alpha Server Cluster (two AS 4100 with a MC link and a 2x2 HSZ70 based disk arrays).

Re:Heh!

[x-caiver]

Microsoft's freebies were limited to an Opteron-specific Windows 2003 Cluster Edition beta and a cookie.

The beta worked on both the AMD and the Intel x64 hardware.

By comparison, many others had booklets on what their products did, papers

There were papers on what the product was. I got a little USB memory key, it was only like 32 megs so not super useful, but free, that had some documents on it. I'll assume that they were smart enough to print those documents out to hand out after the USB keys ran out.

working demos (the molecular modeler with forced feedback was amazing)

Right next to the booth with the 'cookie' were stations with somewhere between 5 and 8 different companies showing off their software sitting on top of the CCS. There were fluid dynamics, car crash simulation, and a couple others. Some of the demos were kinda lame so I didn't pay much attention to them, and a couple of the monitors were showing the same software doing different things, so I'm not sure the exact count.

and some highly knowledgeable geeks to answer detailed technical questions.

There were developers and feature PMs (the guys who write the specs that the devs implement) standing around. I bet that they had a passing knowledge of how their system worked.

I didn't stick around to watch the MS presentation so I can't comment on that. You are right that there were a bunch of other booths that were absolute disasters though.

You said "clippy"

[Anonymous Coward]

I mean, really.

Most programmers who use HPC

[Epeeist]

are not going to be bothered about pointy-clicky interfaces. They will be running highly complex, custom programs that use standard parallelisation libraries. If they want graphics it will be in terms of rendered output.

The last thing they need is a pointy-clicky interface chewing up cycles that would be better used for computation.

Re:Most programmers who use HPC

[Anonymous Coward]

They will be running highly complex, custom programs that use standard parallelisation libraries. If they want graphics it will be in terms of rendered output.

The last thing they need is a pointy-clicky interface chewing up cycles that would be better used for computation.

Indeed. In fact, if one wants the pointy-clicky interface, it will be on the front-end workstation. And by using something standard like MPI to talk to the back-end cluster you couldn't care less if that backend runs on windowsor linux and it's a room of servers or a rack of PS2s.

OTOH, the thought that makes my skin crawl is small edu clusters that more often than not are not well-administered and ripe for hacking. The several cases I know have all been breached into at least once. MS is sure to 'improve' on this with their a-monkey-can-admin-this solution, yeah right! Oh, and of course you need to keep at least the access route up to date, so better put some planned downtimes on patch Tuesday. Productivity, here we come. Yipeee!

Re:Most programmers who use HPC

[nixascyborg]

Thing is; 'most' programmers wouldn't know what HPC, let online how to program for it. MS wants companies to believe that this kind of programmer (or admin) can set-up, program against and run HPC clusters with little training other than Win2003 & VS.NET 2003/2005. If this is true or not is irrelevant; as long as company CTO's/CIO's/CEO's believe it, no one cares about the real technical merit of the statement.
And they do, because they are usually managers who use Word and PPT sometimes and play golf with one of Ballmers' boys; dunno if this is true worldwide, but it is here in the Netherlands. And yes I know this from experience, not from reading it in a blog.
The MS marketing story to managers that are not indept technical is a very strong one; choose Windows, you'll pay, for instance 20 times $500 and use your current Win2003 admins for installing/running and your current devvers or *any* Indian $3/hour company to write your software, versus; pay nothing for the licenses, hire new, hard to get Linux admins for $3000/month, hire even more hard to get HPC / C programmer (also hard to come by when outsourcing...) for the Linux variety. Bottom line; pay, as company, Windows; 10k + 20k computers one time extra for your environment, Linux: pay 20k computers + 40k/year + x * 100k for development.
True or not, this is a strong salespitch which does work on the golfcourse and a lot of companies needing this will go for it, as they have done for other MS technologies which make no sense (embedded? webservers? datacenters? databases? storage software? ... Linux is 'better' than Windows in most cases, but still those markets grow for MS...).

Re:Most programmers who use HPC

[Anonymous Coward]

You obviously don't realise that Windows Server is capable of running in a headless mode with 100% remote administration. The usable pointy-clicky interface - if there even is any - will run on a remote client machine. Not to mention that many remote administration tasks can also be performed from the command-line - you don't always even need to point or click.

So your basic argument is this: you (and by extension, all other software developers) are pompous and arrogant; then you make some shit up; and you also hate usable remote administration. Thanks, Slashdot! I guess that's why I don't visit very often anymore.

BTW, I'm not sure how bad Linux is these days (X was pretty god-damn awful last time I touched it), but UIs have been around on servers since the days of Novell and OS/2. And simply drawing the UI in Windows doesn't take a noticable quantity of CPU cycles - the "it's expensive" argument is, quite frankly, bullshit. Even decoding video is ridiculously cheap - 2% CPU usage for a 640x480 Xvid clip - while HD video (encoded in Quicktime) is about 20% of a 3GHz machine. So unless you have video wallpaper on your server, there really isn't a problem having a UI for administration.

High-Performance Linux

[jd]

If OpenMOSIX is compiled into the kernel, the total effort required to set up a Linux cluster is virtually nil - you need to tell it what nodes there are in the cluster and it transparently takes care of the rest. A home user with ZERO programming experience but has two or more computers, a hub and a working knowledge of what an IP address is can configure a rudimentary cluster in under five minutes. It may not be optimal, it'll rely totally on OpenMOSIX to do the process migration, and without any apps that can take advantage of it, it would be a little pointless, but it could be done. It requires no expert knowledge or significant intelligence. If you can operate VI, you can operate a cluster at that level.

Difficulty, therefore, is NOT a significant factor in all of this. Ok, what about expense? Well, you're right that Linux is free. So is OpenMOSIX, OpenMPI (and many other MPI implementations), PVM (another messaging library), Lustre (a very high-performance network file system), many scientific and mathematical applications for clusters, etc. There are clustering patches for PoVRay, and it's always possible to write a script to have multiple machines render parts of images anyway. I'm sure there are other applications out there that I'm not thinking of right now, and it's only a matter of time before more "mundane" applications can take advantage of clustered environments. They already do, on Plan 9, to some degree. Oh, Plan 9 is also free.

Cost would appear not to be a major problem either, then. Optimizing is the only thing that is in any way difficult, and a GUI system that doesn't let you get to the really fine detail won't help there. More time, effort and money is spent on optimizing than on anything else, and I simply can't see any possible way that an OS that is designed for ease-of-use by hiding the intricacies can in any way help in that.

Re:High-Performance Linux

[Bert64]

Openmosix doesn't even require much in the way of program compatibility... Any program that support multiple threads (ie, multiple cpus) will work with it.

Cautiously optimistic

[hpcanswers]

I think Microsoft's reason for pushing into HPC is to provide better software development tools for clusters. Can you imagine being able to program in VB.net instead of C99? After all, physicists are there to do science, not write code. Plus, MATLAB (Distributed Computing Toolbox) and Mathematica (gridMathematica) will both be available for Windows CCS, and I imagine Star-P may be out before too long. All in all, I'm cautiously optimistic about getting better development environments available for supercomputing. Of course there is still the concern about license costs and the resource-hogging GUI.

I blogged about these topics a while back, both MS in HPC and better programming tools for supercomputing:

http://hpcanswers.com/plog/index.php?op=ViewArticl e&articleId=27&blogId=1
http://hpcanswers.com/plog/index.php?op=ViewArticl e&articleId=25&blogId=1

Re:Cautiously optimistic

[san]

Your comment should be modded funny, not insightful.

The only reason you need a HPC cluster is -- indeed -- High Performance Computing. That means you're going to use as many cycles (or messages passed) as you can get from that $50K+ cluster you've just bought. This easily precludes anything but a fast compiled language like Fortran or C (and no, Java with JIT or .net are *not* fast enough. I've tried).

The comment about domain specific languages in your blog for HPC purposes is true: Fortran is exactly that.

Re:Cautiously optimistic

[Trelane]

After all, physicists are there to do science, not write code.

That depends entirely on the physicist. For some clusterjockey physicists, writing code (usually c++ in my experience) is part and parcel with doing the science, because matlab and mathematica (and maple and well, for some people maybe even IDL) are great for some domains, but not all, and c or c++ will get you to those other domains.

Plus, MATLAB (Distributed Computing Toolbox) and Mathematica (gridMathematica) will both be available for Windows CCS, and I imagine Star-P may be out before too long.

They already exist for existing nix/nux based cluster solutions, save for "star-p" which I've never heard of. They're still Matlab and Mathematica, with all of the tradeoffs that go with each.

Re:Cautiously optimistic

[calyst]

You can not be serious. Most of these scientist guys use Fortran 77, and will not even consider using a RAD tool for coding.Let alone the fact that they already know how to crunch code with built in multithreading for their unix cluster. And of course no university will buy 100+ licences of windows when their admin can produce a unix/linux cluster easily....

Re:Cautiously optimistic

[CrankyOldBastard]

Look at the "quality" of typical microsoft code, and you think that they will give us tools to make developing parallel code easier?

I've yet to find anything close to Occam or fortran suitable for parallel computing and let's face it, except for a few classes of apps (such as nimrod searches), if you stuff it up it's a total bitch to debug.

I find the idea that Microsoft will give us neat tools for this stuff as likely as using VSS for a long time and not having your code corrupted. After all, writing really efficient || code for apps that require different compute nodes doing different things can be really really hard. that's one really fast equals two really hards.

Re:Cautiously optimistic

[Anonymous Coward]

"After all, physicists are there to do science, not write code."

Well, yeah, but don't forget you're talking about people who find LaTeX preferable to OpenOffice or Word.

Re:Cautiously optimistic

[SlashSquatch]

Most of these scientist guys use Fortran 77

In my experience, the above statement is apt.
With Fortran 90 vectorization, staging of mathematics in Matlab, however handy, is almost moot.
Most labs can bring in a first year grad student or senior undergrad and have them code up the experiment they dreamed last night in one day with f90. The next couple days they debug it. It is then assimilated into the main package within a summer. Now, those great open source guys have brought g95 to the table. Linux and UNIX rules in the parallel computing environments. It could be dethroned quite easily, but the usurper better be badass. The main hurdle to overcome is setting up a seamless environment that many users can crunch and offload tons of data. Make a couple small mistakes and the system blows up in your face. Administrators want to know why the quarter, half or full million dollar box won't work. (In some cases, such as petroleum exploration or the accelerators, you are losing that amount of money on a weekly basis if the system is down.) Angry scientists want to know where their hard fought data is. Compound that with the guys who work on this box have their main gig of handling a bunch of windows patches and mailservers. It's not an easy job.

Re:Cautiously optimistic

[AlanWay]

Can you imagine being able to program in VB.net instead of C99?

Of course after writing it in VB.net you'd need a HPC to run it!

Mind you, perhaps they released Compute Cluster Server so people would have something to run Longtooth on?

Re:Cautiously optimistic

[Anonymous Coward]

"Can you imagine being able to program in VB.net instead of C99? After all, physicists are there to do science, not write code. "

If the requirement is to do the research and not code then neither C nor VB.net are what should be used, but rather a language
of a much higher level than either (or Fortran) that can then be targeted, via very advanced compilers, onto a variety of
platforms. Fortran standards for Formula Translator, but is too low level. What is needed is something that lives up to the
name.

Your comments regarding Matlab are nearer the mark, but Windows is not required for Matlab.

Dooom

[MrPsycho]

Ahh I am haunted by the vision of thousands of BSODs running in perfect parallel.

Re:Absurdity can be profitable

[drsmithy]

Something that the majority of Slashdot readers seem not to understand (and with justification) is that purchasing decisions are not rational.

No, something that the majority of Slashdot readers do not understand is that purchasing decisions are rarely made on the basis of up-front component costs.

Uh, did anyone notice the last few years?

[Goodgerster]

2006. It's 2006.

I just can't see any hope for MS

[10Ghz]

I made this comment on Ars Technica, and I'll just repeat it here as well.

I just can't see any future in this. There are few things working against MS here:

a) Price. In the large-scale, the price they are asking would mean lesser nodes. Instead of paying for Windows, the customers could just use Linux and add extra nodes to the cluster.

b) source. Yes it does matter. In markets like this, the people running the cluster do fiddle with things in order to make it go faster. They can't do that with Windows.

c) Ease of use. Well, the people who make clusters are usually not morons, so I don't really see any real need for "point 'n click" GUI for creating clusters. And maybe that GUI could impose a bit more overhead to the system? And creating Linux-clusters is relatively easy.

d) Momentum. Linux has companies like SGI, Cray, IBM and others using and improving it. And there are universities involved as well. Those companies really know Linux and they REALLY know HPC. Microsoft has no real know-how regarding HPC.

e) Familiarity. This time, people know Linux. MS is trying to beat an entrenched competitor. MS has succeeded in doing this before, but they did it by undercutting the competition. This time they are competing against something that is free. And their competitor has the advantages mentioned in A, B, C and D, all of which matter to the target-audience.

Re:I just can't see any hope for MS

[Anonymous Coward]

You're all completely wrong. The market for this windows is here and MS comes with enough power to seize it.

a) Price. In the large-scale, the price they are asking would mean lesser nodes. Instead of paying for Windows, the customers could just use Linux and add extra nodes to the cluster.

When you pay several thousands dollars per node to run your solver (try to find license fees of Abaqus, Moldflow, Nastran...) paying $500 per node is not a problem. And anyway today if you build a professional Linux cluster you will probably buy support for the distro.

b) source. Yes it does matter. In markets like this, the people running the cluster do fiddle with things in order to make it go faster. They can't do that with Windows.

People who want to use proprietary solvers don't care, they want to use the exact version that is supported by the ISV.

c) Ease of use. Well, the people who make clusters are usually not morons, so I don't really see any real need for "point 'n click" GUI for creating clusters. And maybe that GUI could impose a bit more overhead to the system? And creating Linux-clusters is relatively easy.

I don't understand this, what is your conception of a cluster? Users of cluster are not computer scientist and administrators can be windows administrators.

An exemple : A small automotive part manufacturer has a development branch who use mechanics engineers and a guy whose job is to administrate a bunch of windows workstations, 2 printers and a file server. They need a cluster to run some simulations, do you think they would absolutely want a Linux cluster?

d) Momentum. Linux has companies like SGI, Cray, IBM and others using and improving it. And there are universities involved as well. Those companies really know Linux and they REALLY know HPC. Microsoft has no real know-how regarding HPC.

SGI is surely a very successfull example... Microsoft already has the proprietary solvers companies. MS has hired the right people over the last few years.

e) Familiarity. This time, people know Linux.

Maybe but there are still people that today only use Windows workstations and if they consider to buy a cluster this argument will not stand.

Windows CCS doesn't aim for the big clusters of the top500, they don't aim for the small cluster in universities or laboratories. They aim for the middle-range cluster of the R&D offices in big companies, right where the money is.

The question is not "will it succeed?" but "how far will it get?"

no big deal

[excelsior_gr]

I'm working at a university in Germany and we already have a small Windows cluster. However, this is more for research purposes, e.g. to answer questions like "now let's see how this would run on the Windows machine...". The clusters have AMD Opteron or SUN UltraSPARC processors. Check these links out:

http://www.rz.rwth-aachen.de/computing/events/2006 /winhpc/index.php (German)
http://www.rz.rwth-aachen.de/computing/hpc/index_e .php (Egnlish and Gernam)

At a recent workshop the Microsoft guy explicitly said that they know that they don't have the best product on the market and that they are not really trying to compete with the UNIX/Linux systems. They just want to be present in this computing field as well. In the scientific comunity there are people that would use Windows machines as well, simply because they (we) don't care. All that scientists need is the results of our calculations. Our desktops run various operating systems, according to each person's taste. But in the end we all give the guys at the Computing Centre thousands of lines of serial Fortran or C code that they have to parallelize (OpenMP and MPI) and then run. We can also use a cluster remotely, but in this case it doesn't matter which OS you are running. Depending on a person's field of geekness (e.g. chemistry vs. computers), one can either love a command line interface (because one gets to use his/her brain), or hate it (because it takes the focus off the chemistry). Nevertheless, we all know that when one has some serious programming to do, the UNIX systems really rock because of all the cool tools that are available (especially for optimizing and paralellizing code).

By the way, when the article talks about expence, it meens the expense of the hardware, that has been really expensive until now. The cost of the OS, of course, would add to that though.

Re:no big deal

[Slashcrap]

I'm working at a university in Germany and we already have a small Windows cluster. However, this is more for research purposes, e.g. to answer questions like "now let's see how this would run on the Windows machine...". The clusters have AMD Opteron or SUN UltraSPARC processors.

Windows on UltraSPARC?

Stoppped. Reading. Right. There.

Re:no big deal

[excelsior_gr]

hehe,

To tell you the truth, I don't know what kind of CPUs the Windows machines feature. Check the links for more detail.

Excelsior

Re:Absurdity can be profitable

[hackstraw]

...emphasise the fact that purchasing is based on emotion, not rationality. Some 80-90% of all sales are emotion-driven and then sometimes post-facto justified by selectively picking facts.

Maybe with things like cars and clothes, but clusters are merely machines to crunch numbers. Kinda like a big calculator, and little emotion goes into designing and using them. Its bang/buck. Thats it.

Microsoft _may_ be able to sell this HPC edition to some PHB out of emotion who is completely clueless and has clueless admins as well, but an OS has little to do with an HPC system. In fact, the less of the OS the better. Most of the time, HPC apps are in user land. The OS does basic memory management and I/O, but that is it.

Most all clusters are Linux. Why? Its good and cheap. You don't need the scalability and robustness of say Solaris, because you (typically, almost 100% of the time) only have one thread per processor. Yes, I know with large SMP machines, the OS can and does matter, but those rarely have the bang/buck ratio of clusters. The two big guys that have done this over the years (large SMP/NUMA/NUMAcc, etc) are SGI and Cray. And both of those companies are hard for cash right now. IBM probably does not make money, or much money off of their large number crunching systems, but they are probably viewed as RND, not a "for profit" good or service (I could be entirely wrong here regarding IBM, but thats my hunch).

I don't know what Microsoft is doing with this product. Like someone else said, its probably just a "me too!" thing. In looking at their "details", they do not mention using desktop machines at night. The is a BIG miss by them, because that would be one of the only things that could even make this a marketable item for an already primarily MS outfit.

The more I think about this, the more silly this sounds. Yeah, I'm an anti-MS guy, but I try to give them the benefit of the doubt, but this product seems completely worthless. Actually, now that I learned that this is an only 64bit offering, I believe this is a way for MS to sell a product for beta/stress testing of their 64bit server offerings.

To close this post, from the FAQ:

Q. How does a Windows-based compute cluster compare with a cluster running UNIX or Linux?
A. There is little substantive difference, but UNIX-based solutions should be fully ported to Windows to realize the full benefits of the Windows operating system. There are several differences between UNIX-based operating systems and Windows. For example, I/O operations and threading are different on UNIX-based systems than they are on Windows. I/O intensive applications will benefit from using Windows native I/O APIs rather than UNIX style I/O APIs.

Emphasis mine. The second bolded part is important. That porting is expensive and time consuming. Especially when its common for codes to be 30+ years old and designed for UNIX systems. Sounds like vendor lockin to me. Wow, typical Microsoft.

What it means for devs

[Anonymous Coward]

In interesting article about what this might mean for Windows Application Developers can be found here:

http://dndj.sys-con.com/read/45912.htm/

Is there really any need or reason for this?

[multimediavt]

Absolutely not, especially when you also add in the time of application development with MPI. Why on earth would you spend $10,000.00USD for an OS to run a 20 node cluster when your real costs are in the porting and development of software to run on the cluster, not the cluster itself? MS has lost their rack trying to compete in this space! If big business is the target market for this, they're dumber than I thought. Any company CIO or manager that invests in this OS over a free Linux alternative should have their ass removed. The costs for a Windows-based cluster (TCO) would be astronomical when you include application development and porting time ON TOP of the initial cost of hardware and OS licensing! Oh wait, but MS will "support it", right? LMFAO! Sure, ok, they have no idea how much of a support nightmare they are walking into. They are selling an OS, not a turnkey cluster solution and will have no idea the minute details of a customer's cluster setup. They will have to take weeks to diagnose and fix user problems. Also, I can't see the turnkey providers, like Dell and HP, touching this with a ten meter cattle prod! They are having enough fun supporting their Linux installations, that are a lot more mature than this new Windows thingy. I don't see this going much of anywhere.

Automatic Updates

[troylanes]

and virus scans can now be done in O(ln(n)) time!

Wiping the MS tax will double my performance

[leonbrooks]

The "on special with new OEM machine" price of Win2k3 Pro from a typical local wholesaler is less than the price of an extra CPU and a motherboard to add it to. Half as much again for a non-OEM "upgrade" version. I shudder to imagine what an HPC version will cost.

Put another way: I can run twice as many CPUs and save money if I avoid paying the Microsoft tax. I presume that the win is even larger if the licence says "per CPU".

I can probably run four CPUs and twice the RAM for less $$$, given the real HPC price. To say nothing of the absence of viruses, spyware, adware and other interfering, performance-boogering "bonuses" which regularly sing in tune with MS software.

If I also add real reliability figures and "necessary" software licence costs, I could probably shoot for eight CPUs and fourfold as much RAM for less $$$. Then I get to factor in real downtime costs, plus audit fees etc and get much more ambitious.

Oh, yes... and shoving 8x the CPUs and 4x the RAM into just 12% of the same rack-space as the proposed Win machines must represent even greater savings in location and wiring costs. Not to mention the extra foreign-exchange wins and so forth if I dress the servers as penguins instead of mindless Redmond-bound money-suckers.

OK, so saving money and stress while improving yield is simple and straightforward: why not do it?

The Win2k3 5-CAL non-HPC version gets me...

[leonbrooks]

...motherboard, 4 x CPUs, 2 x RAM and a digital still camera, plus I save over AUD$100.00 per system if I don't buy it but install Linux instead.

For the HPC version, I could probably score a DVD camera and a friggin' partridge in a pear tree as well. FOR REFUSING TO BUY W2k3

<SRACASM>I wish all decisions were so hard.</SARCASM>

So can we expect to see...

[Anonymous Coward]

...the "My Cluster" icon on the desktop?

Re:So can we expect to see...

[x-caiver]

That wouldn't be as likely as seeing a Tools:Compute this on a Cluster of File:Offload to Cluster menu option in your favorite Windows application.

DUPE!

[x2A]

MS actually /did/ release it 3 years ago, but you know how slashdot is for posting old/dupe stories ;-)

MS and serious engineers

[asky]

I think MS has a lot more in-roads to technical computing that slashdotters realize.

While working at NASA Ames, I found a lot of technical work on Windows by people who are mechanical, electrical, and aerospace engineers. Applications included: structural/mechanical CAD for fabrication, programs to drive test and measurement equipment, MATLAB to derive simulation models, VxWorks embedded RTOS development.

The universal applications that I saw every engineer use were MS Word and PowerPoint. (LaTeX and troff are largely a lost art; even FrameMaker if used by engineers was running on Windows.) This meant that virtually every serious engineer had a Windows box, and possibly a UNIX box for specialized work as well. Most of these guys understand Laplace transforms, but not regular expressions. They were far more likely to use MATLAB than Perl.

There are, of course, serious software developers there who don't know Laplace transforms, and understand UNIX tools and open source. But this is a different crowd from engineers doing analytical work. (NASA Ames also has a serious supercomputing operation, e.g., the Columbia cluster of 10,240 Linux nodes built by SGI. When you have flow models like protrusions between thermal tiles, this is where you go.)

With budgets really tight and reductions in headcount, they stretch dollars as far as they will go; which means, if it will run on Windows, it's hard to justify another box. Furthermore, system support is outsourced to a department whose sole purpose is to keep computing alive, which means a very limited number of Windows and Mac OS configurations.

Having interacted with engineers involved in aerospace in other parts of the United States, the stuff I saw at NASA Ames seems pretty typical. (When I mentioned this to a Hubble astronomer, he was completely stunned.)

Now I don't expect Windows Compute Clusters in NASA anytime soon. But some engineering software vendor is going to decide that they can extend their product line by bringing compute power to the individual engineer through this mechanism. At that point, a hybrid solution of Windows desktop and Linux compute servers is going to be hard to justify, particularly if it requires additional department resource to make it work.

Re:MS and serious engineers

[Anonymous Coward]

From http://www.microsoft.com/presspass/press/2005/dec0 5/12-09SmithBurtonPR.mspx

REDMOND, Wash. -- Dec. 9, 2005 -- Microsoft Corp. today announced that Burton J. Smith is joining the company as a technical fellow. Smith joins Microsoft as a proven leader in the computer industry, serving most recently as chief scientist and member of the board of directors for Cray Inc. Smith will focus on working with existing groups within Microsoft to help expand the company's efforts in the areas of parallel and high-performance computing. He will report directly to Craig Mundie, chief technical officer and senior vice president for Advanced Strategies and Policy.

Re:MS and serious engineers

[x-caiver]

Related to that post, one thing to remember is: there are smart people at Microsoft.

Just because some products suck (Bob), some features are super lame to technically-inclined people (Clippy), and some decisions were bone-headed (Outlook running macros embedded in random emails), doesn't mean everyone at the entire company is stupid. There are some absolutely brilliant people, a bunch of smart people, some average people, and, yes, even some fools - but that's the same with any company larger then, oh say, 20 people.

Microsoft has made good products, and they've had some duds. I use a Microsoft Intellimouse Explorer on my Mac because i think it is the best 5 button non-gaming mouse that I've used, but I don't use Microsoft Word on my Mac. Pick the right tool for the job, based on the parameters of that particular job and you'll be set. Don't instantly discount Microsoft products just because you think every single person at the company was personally responsible for Code Red.

A WHOLE 4?!

[Anonymous Coward]

"Windows Compute Cluster Server 2003 supports up to four processors per server."

Wow! Four processors in a single box! That's amazing! This will definitely smash Linux!

Re:A WHOLE 4?!

[x-caiver]

Windows Server 2003 Enterprise x64 Edition supports 8 in a box, so if you happened to have 8 proc boxes you could just upgrade the OS.

Think about making a cluster out of blade servers, etc. Those aren't up to 16+ procs yet, so either the Compute Cluster Edition or the Enterprise Edition are fine in this case. You're talking about a totally different type of computer when you're talking 16+ procs. (8-64 procs requires 2003 Datacenter Edition which is an OEM product only available in conjunction with a certified DTC system)

Some informed views on WCC

[Anonymous Coward]

See Scalability.org and the referenced HPCWire. The people writing these have a clue, and are in this market. Ask them what they think.

At least it follows standards....

[ShyGuy91284]

They have at least decided to implement the MPI (Message Passing Interface) standard for this cluster system. Hopefully it will be somewhat compatible with other MPI implementaitons like the C++ one I have used (I also used Java, but so far, I'm not impressed by Java MPI implementations).

it's 2006...

[ltwally]

"...on Friday Microsoft released Windows Compute Cluster Server 2003."

I just wonder if MS realises that it's the year 2006....

accept the cookie

[bmidgley]

At the supercomputing 2005 conference, Microsoft made the unfortunate decision to hand out cookies with "Cluster Server 2003" on them. We enjoyed joking about what a mistake it is to accept cookies from Microsoft. Not until later did I think about the date! Nearly 2006 and they were giving out food items with "2003" on them!

How fast?

[codepunk]

How fast will spybot search and destroy run on it.?

Seriously by the time you add virus scanning, spybot removal tools and all the other crapware you need just to keep one running half your cluster clock cycles are gone.

When you sip from the cup of Microsoft

[LazManico]

Licensing costs += $$$

What is the point ?

[Salsaman]

If you are running a 1,000 node cluster, do you really want each node to be wasting resources running a GUI ?

If the nodes aren't running a GUI, then would you rather be running DOS or bash on them ?

Actually they don't but

[crovira]

now that super computing has been turned into clustering and there are lots of people doing it (like it hit >$x billion,) it has apppeared on microsoft's radar.

Unfortunately for Microsoft, the terrain's already covered by Linux and those systems are a moving target with cost-benefit lines that Microsoft CAN'T possibly over take. (The software is $-free and open source and the users WANT collaboration.)

Its a technological death trap for Microsoft. (I can just hear the SNAP. :-)

20 machines need a special cluster OS? Nah...

[level_headed_midwest]

The university that I attend, the University of Missouri, has two new clusters for their bioinfomatics analyses. One is an SGI Altix 3700 Bx2 that has 64 Itanic 2 processors and 128GB RAM. It runs Suse Linux Eneterprise Server 9.0 Server, which is just the Novell-supported version of plain old SuSE Linux. The other is a 128-node Dell Xeon DC+DP cluster with 512 cores and 640GB RAM. It runs Rocks Linux 4.0, which is a specially-tweaked cluster version. The university is SO pro-Microsoft that it's not even funny, and yet they still use Linux or FBSD on all of their servers and HPCs except for the 36 Windows 2003 servers that run Exchange (there's one server per every 1000 e-mail addresses.) I think that says something about the limited viability of Windows in a server role except to run Microsoft programs and very small terminal and file servers.

Re:20 machines need a special cluster OS? Nah...

[Anonymous Coward]

Probably a bit of an oversimplification to say that SLES 9 running on an Altix is "just the Novell-supported version of plain old SuSE Linux", as SGI actually develops a whole series of "big iron" extensions to the distro (they call their "ProPack") that handle such things as cpusets (nodeboard-level cpu/memory affinity and compartmentalization) on that specialized hardware.

Re:20 machines need a special cluster OS? Nah...

[level_headed_midwest]

The SGI machines did have the ProPack installed, you are correct.

Re:Absurdity can be profitable

[iluvcapra]

should be fully ported to Windows to realize the full benefits of the Windows operating system

I got the impression from TFA that the major feature HPC Edition had to offer was its handy and clean interface with Visual Studio, allowing all the CLI masses to write software for a compute cluster all their own (without having to deal with learning a different platform).

Would this have any effect on the HPC market? I can't see people with existing installations+software biting, but I can see it tempting to businesses building new installations and wanting to minimize training costs, etc, even if what they're buying won't really give you a full "H" in HPC. If a business IT department has passed the dictum: "This is a Windows only environment," and a manager in that business needs a compute cluster, he can either pick a fight with his IT department or bend over and take the Windows HPC software (which MS will give them for peanuts this first time around).

Unrelated, Near future prediction:

1. MS wil require the cluster edition in order to run distributed MSSQL
2. The number of "clusters" in the world will go up stupendously (MS will define this statistic as counting all machines which run a clustering OS, regardless of the application)
3. Instantly 80% of all "clusters" will be running Windows HPC
4. This fact will appear in marketing everywhere
5. Profit

Give me the benchmarks

[fluor2]

Give me the benchmarks please, and I'll consider it. I guess some mathlab and such would be worth comparing.

I dunno

[KwKSilver]

This is aimed at unspecified big businesses who may need clustered computing.

Clippy: "Looks like you are tring to calculate how much you can skim off of the sales tax collections without attracting attention ... May I suggest..." OR "Hey, those old fossils don't deserve to have retirement or savings accounts anyway, let me show you how to temporarily divert a point or two of of their interst and dividends in a way that can never be found out, and will alows for much better payoff on executive stock options."

I had hoped you were kidding...

[KwKSilver]

But according to Argonne's own site they use

The division's computing environment, consisting of nearly 2,000 different computers, is based on a network of diverse workstations including Linux, Windows

God help us....

Re:MS coders and intelligence

[KwKSilver]

I keep seeing the statement that MS has smart people, as if the perception they don't was their only problem. IMHO, it isn't. As much as I hate to credit MS for anything, Excel is a superior spreadsheet, possibly the best. MS mice are good, too. What MS lacks more than anything else is a history of integrity. Shipping date over stability or safety. MS has been sued more over IP misappropriation than anyone else I can think of. The term vaporware may not have been coined for MS, but they are the epitome of shrewd use of it. MS would be less dangerous if they weren't so smart: if smarts + unlimited greed + 0 ehics doesn't add up to MS, I don't know what does.

Filling in the blanks...

[SteeldrivingJon]

Since everything on the netzilla is calledzilla zillazilla these days, I'll fill in the details for the grandparent commenter.

But by 1989, Richard Crandall, now Distinguished Scientist at Apple (and once my roommate at Reed College), started networking NeXT computers to find, factor, and test gargantuan prime numbers.

"Community supercomputing occurred to me one day at NeXT engineering headquarters," Crandall recalls. "I thought we ought to make these machines do what they were designed to do, which is to work when we humans are not working. Machines have no business sleeping."

Crandall installed software that allowed idle NeXT machines to perform computations, combining their efforts across the network. He called this software Godzilla, but after a legal inquiry from the company that owned the rights to the movie character, he renamed it Zilla. Crandall put Zilla to work on huge prime numbers, which are crucial in cryptography. It was then used to test a new encryption scheme at NeXT - a scheme now employed at Apple, which acquired NeXT. In 1991, Zilla won the Computerworld Smithsonian award for science.

Later, Crandall and several colleagues used distributed processing to complete the deepest computation ever performed, asking the question: Is the 24th Fermat number (which has more than 5 million digits) prime? "It took 10**17 machine operations - 100 quadrillion," Crandall says proudly. "With that level of computational effort, you can create a full-length movie. In fact, that's about the same number of operations Pixar required to render A Bug's Life."

Re:Filling in the blanks...

[Anonymous Coward]

Wow. So Apple/NeXT invented clustering. DEC must be so pissed.

Are you retarded?

[Ayanami Rei]

Booting from SAN doesn't require a local disk.
Instead it requires:
1) An expensive fiber adapter
2) Fiber switch
3) SAN storage equipment... filled with a substantial number of drives, at great cost and power consumption.

Which Windows will turn around and treat like a local disk. Blargh.
Let me know when you figure out how to boot Windows off a single read-only LUN or from a LAN.

2003?

[solomonrex]

Why look at the calendar? I'm eagerly awaiting Windows Vista 2000! A very good vintage for MS!