Lights On But No One Home At Sun Grid

cygnusx writes "The Register reveals that Sun's pay-for-use grid computing services hasn't picked up a single customer yet." From the article: "The missing customers prove quite shocking when you consider that utility computing users must agree to be named in marketing programs as part of their contract with Sun - a fact learned by The Register and confirmed by a Sun spokeswoman. More than one year since it first started hyping the 'pay-for-use grid computing services' Sun is still weeks away from presenting a customer to the public. The program has proved much tougher to sell that Sun ever imagined."

Comments

It's entirely SUN's own fault

[Work Account]

You don't embark on a large project of ANY kind without at least securing a customer or two during the development process.

Unless of course you're doing something with free software like Bittorrent where you don't need to money and everything else is cost neglible.

Re:It's entirely SUN's own fault

[Jeff DeMaagd]

The problem with that comment is that free software development is often done as a hobby, not as a paying job. Sometimes it can be used to make a name for yourself too, though that's a pretty elusive goal.

Re:It's entirely SUN's own fault

[Anonymous Coward]

The average software developer could survive without eating for a very long time!

And probably be healthier too.

Re:It's entirely SUN's own fault

[eln]

When you develop a project like this, you usually need to have enough money to run it profit-free (and even revenue-free) for some period of time so you don't need to have customers commit to buying vapor from you in order for the thing to work.

That being said, however, you still need to have done some realistic market research. They should have at least contacted some friendly organizations (current customers mainly) to gauge actual interest in this thing beyond just the "that sounds cool" stage. The larger the financial risk involved in the project, the more market research needs to be conducted to mitigate that risk.

It sounds almost like someone at Sun got a "really cool" idea, and everyone else at Sun thought it was super cool too, and no one bothered to ask anyone on the outside. Or if they did, they only paid attention to the ones that said it was cool and ignored the others. Or they only asked people if it was cool, and never asked them if they would buy it if it were available.

It seems like Sun badly misread the market here, and I would assume someone in their marketing department is going to have a very bad day in the near future.

Re:It's entirely SUN's own fault

[Anonymous Coward]

You don't embark on a large project of ANY kind without at least securing a customer or two during the development process.

True enough, but requiring your customers to add their name to your advertising campaign? that's silly.

Suggest it, maybe even request it after proven performance, but don't require it.

Re:It's entirely SUN's own fault

[dwandy]

...just when I was starting to miss the good'ol dot-com days...

It's so nice to see that not every company has abandoned the idea of not having a revenue stream...

Re:It's entirely SUN's own fault

[SocietyoftheFist]

Sun misread the market? Well duh, they've been doing that for a few years now. Sun seems to think an awful lot of themselves, and they do make a hell of a UNIX, but they don't drive the market, the market drives the market and they don't seem to realize it.

Re:It's entirely SUN's own fault

[Orasis]

Sure you need to be able to take a loss on it for a while, but the original poster is correct that you need to have customers driving your requirements from day one.

Re:It's entirely SUN's own fault

[Craig Ringer]

Yep... but for all we know they did have customers lined up as it was being built. It wouldn't be the first time that a customer goes on and on and on about how great something will be and how much they're looking forward to it, then when it comes time to hand over the money they "reevaluate their priorities".

Technology companies, especially those customising software for a client, know what this is like all too well.

Re:It's entirely SUN's own fault

[mr_stinky_britches]

I am not even aware of anyone attempting to compete with them for this market. They definately should have done more market research and found at least a few potential customers before they went and dropped big bucks on this project. I know kids in highschool who could have told them that..

in the meantime, maybe they should just let non-profit organizations use it ;) now there's a thought.

Re:It's entirely SUN's own fault

[Anonymous Coward]

I am a Sun Employee

The Register has, to some extent, got it's self mixed up.

The Grid Utility offering currently exists in 2 flavours which are still fairly fluid and are evolving to meet the markets needs.

The first is a 'enter your credit card number on our secure website, submit your job and wait for the results' ('Retail Grid') which has been on limited release to early access customers for a while now. I think the reason there has been little publicity around which customers which use this part of the service is because this model isn't contract based. As I understand it, people signing up on the website do not necessarily have an agreement with Sun over publicity.

The second model (the 'Commercial Grid') is a more tailored customer grid which does involve contracts and engineering development work whereby a customer is expected to return to the grid periodically to use 'their grid environment'.
This service has been in use for many months and although this part of the service *was* slightly delayed, we currently have a significant number of customers and potential customers who are conducting testing and running jobs on the Commercial SunGrid.
One thing we aren't suffering from is a lack of interest,

Also, The Register seems to have forgotten about this: http://www.internetnews.com/ent-news/article.php/3 529891

Re:It's entirely SUN's own fault

[Cromac]

in the meantime, maybe they should just let non-profit organizations use it ;) now there's a thought.

Given their lack of paying customers that's a pretty good thought. That might give them clients they could point to when trying to attract customers showing how well their service actually works. Of course if their service doesn't work then I'm sure they'd rather lock someone into a contract before word gets out. :)

Re:It's entirely SUN's own fault

[owlstead]

"Sure you need to be able to take a loss on it for a while, but the original poster is correct that you need to have customers driving your requirements from day one."

That's like saying that marketing push does not work. For generic products like this it is very unlikely that you get any customers buying it up front - it does not directly solve any problem. The problem here is that there is probably only a niche market. So you will have few people interested in it as well. That makes it hard to do a market push.

Re:It's entirely SUN's own fault

[Usquebaugh]

As long as a couple of other million people are as well.

Re:It's entirely SUN's own fault

[Jimmy The Leper]

Probably not, but without amatures the airplane would never have been invented.

Re:It's entirely SUN's own fault

[RingDev]

Was it? Did the Wright brothers have funding? or was it a completely self funded operation? But I agree with you in principal. We wouldn't have linux with out the amature work of Linus. But at this point, a lot of professional OS code has been added. OS is a great tool to keep competition and innovation in a capitalist economy.

Unfortunatly, it has a negative effect on the value of programming jobs where the product competes against an OS/free (as in beer) product. I like Open Office. It saved me from having to drop a chunk of change on MS Office for my online classes. But if everyone switches to Open Office, or other free (as in beer) products, Microsoft will no longer be able to sell MS:Office. Which means that the MS:Office development staff will likely be let go. You then have a couple hundred highly skilled coders, documenter, managers, analysts, designers, etc flooding the market. Increased supply and decreased demand means that costs drop. Good for businesses, bad for professional programmers. So you wind up with a reduction in the number of professional development houses in the US and technology innovation continues moving overseas.

Got to love the world market, eh?

-Rick

Re:It's entirely SUN's own fault

[Znork]

"Unfortunatly, it has a negative effect on the value of programming jobs"

It's not quite that simple. When you free up resources spent basically reimplementing the same wheel the last decade, you also free up the resources spent by other companies buying that wheel. That means the other companies suddenly have more available resources they can spend on custom software improving their own business. Or they can lower their prices, in the end leading to you getting more value for your paycheck.

In the end, any increase in actual wealth for society as a whole is driven by things getting cheaper to produce. When you eventually approach a zero cost due to the nature of an infinitely reproducible product, you have effectively ended scarcity for that product and there is a permanent increase in wealth for society.

Good for businesses, good for consumers, good for programmers who can move on to new things instead.

Perhaps we'll eventually run out of that many new things to do, but that will mean we've also run out of scarcity. And when there is no scarcity, well, having some more free time would probably do wonders for the average programmers stress levels.

Re:It's entirely SUN's own fault

[jcr]

Did the Wright brothers have funding?

They used family money, mostly from their successful bicyle shop. As it happens, the very well-connected (and taxpayer-funded) Samuel Pierpont Langley failed to get a manned aircraft aloft.

If the US government wanted an airplane, they should have just posted a prize. The fifty grand they gave Langley could have easily resulted in ten or more viable contestants.

-jcr

Re:It's entirely SUN's own fault

[InvisibleSoul]

You don't embark on a large project of ANY kind without at least securing a customer or two during the development process.

Whatever happened to the good ol' "If you build it, they will come" philosophy?

Re:It's entirely SUN's own fault

[Decker-Mage]

Actually this is nothing new with respect to Sun. They've been stuck in the (dumb-)client/(smart-)server mode ever since I can recall and that's going back quite a ways. They keep hoping that the market will come around to their way of thinking but, frankly, I can't see it. The fundamental problems of information/systems security, compliance, and other issues are pretty much insurmountable in my not so humble opinion. Yes, they can be addressed but are you willing to bet the company on it which is what it comes down to when you get down to brass-tacks. I won't. To put it simply, I'm absolutely ruthless about systems security and info-sec here and I get all the security notices, so they can't tell me that they are "safe". Thankfully I don't have to worry about compliance issues, yet, aside from the usual contractual NDA's.

This doesn't even address the price-point issue. Sun set that at $1 per cpu-hour. Where they pulled this figure from I have no idea but looking at it I can't see it used for anything but trivial usage patterns or by enterprises with large IT budgets. Just on the basis of my relative's usage patterns the cost would dwarf their satellite TV bill completely despite having all the channels and they seem to be fairly typical users. That doesn't even begin to address my usage pattern which would easily finance a new computer a month.

Sorry Sun, back to the drawing board. Better yet, talk to IBM which seems to have a clue in marketing grid super-computing. That's the market you really need to play in, if you can compete. Frankly, I don't think you have a prayer there either which is sad. They actually are coming up with some pretty nicely engineered designs of late at decent price-points.

Duh..

[shbazjinkens]

Well, it isn't like computers are so prohibitively expensive that everyone is rushing to use this anyway.

Re:Duh..

[Guspaz]

Exactly. Do some math. I need a task done, and it is going to take 1 million CPU hours. Maybe I need to render a movie or something. I need it all done in 1 month.

Well, I could pay Sun a million dollars for ~1400 CPUs for a month, or I could spend about a million dollars and get 350 dual-processor dual-core Opterons, use them for a month, and then sell them at pretty close to retail, bringing my costs to way under a million dollars.

Or you can keep them and use them for more projects.

Either way, Sun's solution isn't really cheaper than a company doing it yourself. It's more expensive than buying the hardware yourself and paying some people to set it up. This is why when small companies need to render EFX for a movie, they buy up a lot of hardware, use it to make the movie, and then sell it off again. It's cheaper that way. They did this for Riddick.

Re:Duh..

[elmegil]

Sun's solution isn't really cheaper than a company doing it yourself.

Because, after all, those machines are all self-maintaining and configuring.

Re:Duh..

[Zemplar]

You apparently don't understand the concept of grid computing and the expected customer base. Render farms are not, nor have they ever been, the anticpated customers for this service.

Rather, large business needing to run monthly or quarterly Monte Carlo simulations, where you need massive power but only intermittenly, are the targeted customer type. In this case, is would be much cheaper and easier to use Sun's service than try to accomplish the same goals in-house.

Re:Duh..

[LurkerXXX]

Brilliant. You've gotten your bean counters to cut a check and 350 dual-core, dual-processor Opterons are in a truck outside. Where are you going to put those 350 machines? You need some space. Another expense. That's a lot of hardware. Even being up for a month they are going to use a lot of electricity. Hmm, damn, it's hot in the room now. 350 high end dual-cpu machines generate a lot of heat, so now you need to back to the bean counters again and ask for money to buy extra AC units and have the HVAC guys hook those into your building. It's not cheap, but still cheaper than Sun!

Great. Now you need to buy a bunch of racks and equipment to connect those 350 machines togeather. Make that 350+, with that many going, your likely to experience some hardware failures. Your getting to be close personal friends with the bean counter. It's not cheaper, but it's gotta be cheaper than Sun.

Now we're cooking with gas. Now you you just need to hire someone who knows what the hell they are doing to hook them up in a workable cluster, tweak the hell out of it to get anywhere decent performance... hmm. Gotta either hire a body or contract out. HR won't want a body for short term. Ah well, better hire the expensive contractor since this is all going to close down in a month. Hey Mr. Beancounter, I need a contractor that's worked with big clusters before to spend some time here. Ouch, one who is really good, available, and will do it well and quickly is not gonna be cheap. $$$. Oh, and hope they don't botch the job so that it takes hiring another contractor (and more time) to do it right...

Ahh, finally. We've cranked through the job. Excellent Now we just have to go back to our friend Mr Beancounter and have him put 350+ odd machines on Ebay for us, plus the racks, switches, cables. Oh, and the big AC units... hmm, some of this might take quite a bit of time to sell to start recouping the money. Ah well, our beancounter wasn't going to be doing much else for the next few months. Or won't be now anyhow. Oh, plus we need to have someone tear down and box everything up. Probably for shipping to 1000 different places. Hmm, and the HVAC guys will have to come uninstall the extra AC units. But in a few months I'm sure we'll have back part of the costs of those machines!

A bit of a hassle, no? A big expense. You might, just might come out cheaper than if you went with Sun. Then again, you could just cut a check to Sun, know the job will go into their queue, and you will probably have your data back from them in a few days.

Sometimes it's just easier to pay someone who does that stuff for a living.

Re:Duh..

[tchuladdiass]

But, how many of those customers have a bunch of PC's sitting on employee desks that are idle 99% of the time?
Even if the PC's are running windows, it would still be a fairly easy task to push out a colinux image via sms, and have the image automatically join the cluster. As long as the cluster software can account for nodes joining and leaving the network arbitrarily, then it could be a more cost-effective solution.

Re:Duh..

[Anonymous Coward]

Your post is absurd.

Before getting to your post, Sun missed the market. Those who can pay the extravagent costs WANT the computations done in house due to liability, secrets/privacy/control, and esteem/pride issues. Those who really want this service CANNOT pay the hourly cost, iow Sun has established they are too small or they cannot recover costs from catering to those customers. Most high end renderers can't afford the hourly cost; those in finances already have their own resources.

"Now you need to buy a bunch of racks and equipment to connect those 350 machines togeather."

What the hell. Like this is hard if you just sourced 350 x 2 processors and 350 motherboards and 350+ sticks of mem?

And that's overlooking why'd you'd rack these up. You don't need racks to slit holes into a $10 piece of metal to frame these up. The most difficult part is reducing the number of power supplies, which is not that difficult of a problem either. Even sequencing these up is trivial given the ATX spec. A single person could have this up for a pay scale of $10,000 a month.

"Now you you just need to hire someone who knows what the hell they are doing to hook them up in a workable cluster, tweak the hell out of it to get anywhere decent performance... hmm."

This is bullshit. How is this ANY different than compared to running on Sun's system where you have to develop the software to run on their particular (closed, unchangeable) configuration. I would also say that it's easier to do on your cluster, given you can reconfigure it minimally, IF it became necessary and you determined it was a hardware setup issue. Plus, most clustering software will be programmed to maximally configure itself anyways, if not innantely by the nature of the computations being done.

"Where are you going to put those 350 machines?"

Land and space is ridiculously cheap stateside. You stupid enough to put this in the middle of a downtown area or something? Where I am, you can get a 2 story standalone 2,500 square foot commercial building with 3 phase power for under $1,400 a month. Need network space? Go down one state and the real estate is the same plus cheap DS3 connections.

"That's a lot of hardware. Even being up for a month they are going to use a lot of electricity."

Err, no. Certainly not compared to $1,000 an hour or more. Run a self-built cluster for a full day, the savings would immediately offset (compared to paying Sun) 3 months of electricity usage and pay for the whole year in real estate costs.

"Hmm, damn, it's hot in the room now. 350 high end dual-cpu machines generate a lot of heat, so now you need to back to the bean counters again and ask for money to buy extra AC units and have the HVAC guys hook those into your building."

Here in the NE in the US, 9 months out of the year you'll simply need to run fans to vent to the outside. The thermal mass of the whole damn acreage outdoors trumps your AC units. Do not make the mistake that AC is the answer; mass airflow is the answer. Oh, I forgot, you're stuffing your setups into closed little racks and rackmount cases so they choke--my mistake.

"It's not cheap, but still cheaper than Sun!"

It's a HELL of a lot cheaper than Sun if you run computations regularly. Heck, the case in point indicates this to be true because NO ONE has paid for their services. Have you? No. Has anyone else? No. Sun's cluster is a novelty right now.

"We've cranked through the job. Excellent Now we just have to go back to our friend Mr Beancounter and have him put 350+ odd machines on Ebay for us, plus the racks, switches, cables. Oh, and the big AC units... hmm, some of this might take quite a bit of time to sell to start recouping the money."

Gimme a goddamn break! You're probably paying Mr. Beancounter more than you'd pay the IT guy setting this up. Mr. Beancounter would be smart enough to set this up as a division and separate company, which leads to immediate tax breaks such that $125,0

Re:Duh..

[Minna Kirai]

Rather, large business needing to run monthly or quarterly Monte Carlo simulations,

Users like that will fairly quickly decide that if monthly simulations are good, then weekly will be even better. And while they're at it, do 4x more variations on each sim, to get "better" outputs.

So-called "intermittent" jobs have a way of expanding to fill all availible processing time. It happens on desktop PCs, it can happen on enterprise-level sims too. Soon enough, "intermittent" has become "twice-annual maintenance shutdowns".

Re:Duh..

[slavemowgli]

Just like Sun's machines that are used in the grid.

You can't have it both ways. Sure, computers incur running costs like electricity, maintenance and so on, and sure, you have to take that into account to evaluate how much running your own server(s)/mainframe/grid/cluster costs. But then, the same also goes for Sun - and Sun will certainly not just pay that for you because they're nice; they'll add those costs to the price they charge you. Maybe they'll still be cheaper (or maybe not), but if you point out hidden costs in one option, then you should also point out the same hidden costs in the other option, because in the end, it's *always* you who pays for them.

Re:Duh..

[LurkerXXX]

I would dare say you have never run a company, never even looked to comprehend taxes for a business, and worse, never setup a cluster greater than 25 machines or even read an online research paper about clusters.

And I would dare say you didn't even read the post I was replying to where he spacifically stated he would buy it, use it for one month, then sell it. Not

If it meets the need, it stays there to be run the next time or quarter without ANY cost to the company in terms of additional hardware.

That's specifically NOT what the grandparent proposed. Learn to read.

It's a HELL of a lot cheaper than Sun if you run computations regularly.

No doubt. That's why people who do need a cluster regularly buy their own. Sun intends this for folks who don't need it regularly and would rather farm it out than to buy a cluster, run it for a week or two a year, and otherwise have it sit idle. It's not worth it.

Of course it might be to your, with your your nice big rented warehouse, computers set up on old rusty filing cabinets or whatever spare rubbish you can find to pass for deskspace. I'm sure it would come out cheaper for you. I'm just not sure how happy your boss will be when you tell him that your sorry, but it's warm for June, and he might going to have to wait 4 or 5 months till it gets cool enough outside for you to crank up your cluster.

I'm sure someone has put you in charge of a large cluster... They'd deserved what they got I'm sure.

Re:Duh..

[elmegil]

My main point was that your analysis is laughably oversimplified, to the point of adding absolutely nothing to the discussion.

Re:Duh..

[elmegil]

You build your own furniture starting with the trees in your vast yard, don't you? Certainly that's more cost effective than getting someone to do it in bulk...

Re:fp?

[EvilEddie]

Obviously this news article is about as popular as SUN's new program

Secret Projects?

[HTH NE1]

Does this mean that the only reason why someone would want such computing power is because they want to run projects they wouldn't want the public to know about?

Re:Secret Projects?

[xtracto]

Does this mean that the only reason why someone would want such computing power is because they want to run projects they wouldn't want the public to know about?

No.

The basis of their project is that it would be better for companies to buy processing time than to build their own distributed processing network.

Of course it is interesting to see who (in the real world) are those companies?. If we suppose they are some top-notch companies that use a lot of processing power (like stock market companies wanting to run their models) they may preffer (and they may already have) to run their own servers to protect their secrets.

If they are not so big companies with not too much data then they may have enough power with a beowulf cluster of this-and-that.

The main problem I see here is that any company willing to "buy" this power have to ponder at least this two issues:
- They have to give their data and algorithms.
- They have to relay on SUNs servers stability.

Now, I think the theory behind this service is quite good and, I am thinking to use it as an application case for my risk management on multi agent systems thesis. But I hope when I start looking at the test cases there are at least some companies over there using it.

Re:Secret Projects?

[monkeydo]

The Register concludes that because Sun hasn't publicly announced any customer, that there aren't any. But, they based their conclusion on a false premise. Just because the standard agreement for this service contains a publicity clause, doesn't mean that it will remain in whatever contract is actually executed. If $customer came to Sun and said they wanted to use your service, only if they would strike the publicity clause from the contract, Sun may have been willing.

Re:Secret Projects?

[M-G]

This is very likely to be the case. Many copmanies don't want their name used for a supplier's marketing purposes.

However, I still suspect that there aren't that many actual customers. The whole 'utility computing' concept will be another buzzword that gets lots of press in the industry rags, will find a few niches in which to operate, but doesn't really change the world they way it's touted to.

Re:Secret Projects?

[The Plebe]

Yeah, I agree with this one. It's possible there are some stealth customers, but their service pretty much blows. I recently went through a full-blown grid business case and on the technical side what Sun was doing didn't make any sense. They've since said that they were willing to run grid managers from anyone (like DataSynapse) to make up for their own crappy products (which aren't much more than a scheduler and some scripting) but even then the $1/hr/cpu only covers CPU -- you still have to pay extra for network connectivity to their lab, storage, etc. I doubt anyone serious enough to look at this model (e.g. financial services companies) would be able to make a good case at this time, especially if they're already running in-house grids.

I volunteer!

[Anonymous Coward]

Can they compile my Computer Science homework for me?

They never called me back..

[erwin]

I sent an email inquery to them right after it was announced, and no one ever contacted me. I even talked to someone at Sun (a different division), and still never heard from them....

Re:They never called me back..

[AKAImBatman]

It's probably because Sun screwed up the focus. Instead of selling to anyone who will buy, Sun usually sells to only large companies that wave MegaBucks around. What they don't realize is if they got a large number of "small" companies on the N1 system, they may be able to convince the big boys to jump on the bandwagon.

Personally, I see this as being particularly helpful for rendering houses. The cost of running renderings is simply astounding, with many small companies having to purchase grid time from larger rendering houses. If Sun's computing grid is as cost effective as they make it out to be, it could be a boon for these small time 3D shops. At the very least, they'd add more "little guys" to the mix to help Sun sell the N1 to the big guys.

Re:They never called me back..

[Wesley Felter]

Also consider that big companies tend to already have their own grids. Sun Grid sounds like it would mostly benefit smaller customers.

Re:They never called me back..

[Cervantes]

You might have had more success if your email hadn't referenced Beowulf clusters so extensively.

Although the pix of Natalie Portman and hot grits that you attached probably got passed around a bit.

Cost of porting, uncertain future of the service

[UR30]

Who would be willing to commit their resources to running applications on this system, which has no guarantee of existing after a couple of years? Selling computer platforms to customers, or providing a comprehensive ASP-style solution are more straightforward business models. And can Sun guarantee that data and applications will be secure on their grid system?

It could simply be advanced trolling

[Work Account]

Nothing like that even exists. A bored PR person sent out a fake Onion-style news release and the rest is, as they say, history.

You were just one of the biters ;-)

Ha ha

[Moby Cock]

It really tickles my funny bone to see big companies make such big mistakes. I realise that this makes my a very cynical person but I can't help the way I feel. I like it even better when Hollywood makes giant flops. Remember that stupid Alexander the Great movie last year?

Seriously though, why would someone subscribe to this service? Its not like computers are overly expensive anymorew and there is a fairly broad base of expertise to draw upon nowadays for system admin services.

Re:Ha ha

[LilGuy]

Remember any of the retarded movies made last year? Or in the past 5 years for that matter?

I don't. Until I go to the movie store and can't find anything worth watching in the New Releases (because 5 year old movies are still considered new releases half the time).

Seems to me, big industries are much more error-prone than the little guys. They can afford to be.

Re:Ha ha

[EnderWiggin99]

Remember that stupid Alexander the Great movie last year?

No.

Re:Ha ha

[morgan_greywolf]

Seriously though, why would someone subscribe to this service? Its not like computers are overly expensive anymorew and there is a fairly broad base of expertise to draw upon nowadays for system admin services.

Ever manage a grid before? I have. Once you get beyond a few machines and start running enough jobs to fully utilize all that hardware, management becomes a non-trivial task.

Some companies may want to utilize a grid, say for rendering, but they don't have the IT resources to manage such a system. Especially if their rendering needs aren't so great that they need a grid system full-time -- think small CGI studios or architectural firms that use visualization -- they won't be able to afford the IT resources to manage such a system, either.

That's why there exist service bureaus that have large rendering farms available for hire. Only many of them charge much less than $1/cpu-hour.

Re:Ha ha

[tez_h]

Seriously though, why would someone subscribe to this service? Its not like computers are overly expensive anymore...

I think I said this last time this topic came.

Essentially, you get to pay for the exact amount of compute power that you want. If you aggregate the cost of the hardware, maintainence, etc over the actual utilised cycles, you'll find it's probably much higher than Sun's offering, precisely because they have gambled on getting economies of scale.

It seems, for the present anyway, that it has been a bit of an expensive commercial experiment.

-Tez

Re:Ha ha

[Surt]

You would subscribe to this if your small business needed a million cpu hours next week, and all you had was 100 computers.

Re:Ha ha

[lpangelrob]

Heh.

The only reason I remember that movie was because I remember it came out either directly with, or two to three weeks after, Troy, which was arguably a Hollywood success (assuming break-even to be successful).

That was not a good idea — Brad Pitt, and if I recall correctly, the currently-invicible Orlando Bloom, were in Troy. I can't recall the name of a single character in Alexander.

Re:Ha ha

[BewireNomali]

dude, angeline jolie was in Alexander.

MILF.

Re:Ha ha

[thesandtiger]

It really tickles my funny bone to see big companies make such big mistakes. Like, remember that dumb move by Apple where they started offering a product called the "iPod"? Or when Microsoft was really stupid and decided to offer something called "Windows" despite the fact that DOS was doing just fine? And some stupid company called "AMD" is selling processors even though Intel should be good enough for anyone!

Anyone remember that stupid movie "Star Wars" that came out a few years back? Can you believe that anyone was dumb enough to try putting that thing in theaters? Or that stupid television show "Star Trek"?

Companies make big mistakes because they take big risks. Sometimes it works, sometimes it doesn't. Sometimes it takes off like nothing anyone could have possibly expected. Sometimes it smashes into the ground and makes a crater so large it takes the company down with it.

In fact, you seem to be advocating caution. The same kind of caution that had the guy over at HP ask why anyone in the world would want a personal computer. Or that those IBM execs exercised when they made the "safe" bet that the hardware, not the software or operating system, would be the big deal down the line.

Re:Ha ha

[Lucractius]

while its possible to use this for rendering ill repeat again the words of others. Its not meant for people wanting to run rendering. Instead for people that need short term high volume proccessing at itermittent intervals.

Say a small (but lucrative) investment firm with cash to spare but not enough to manage an IT project the size they need for the simulations they need about 25000 hours for every 3 months and that would be sitting idle the rest of the time.

Re:Ha ha

[Moby Cock]

In fact, you seem to be advocating caution. The same kind of caution that had the guy over at HP ask why anyone in the world would want a personal computer.

I am advocating nothing. And I'll be blunt: I think its funny that some guy at HP thought the PC would be a bad idea.

Re:Ha ha

[Blakey Rat]

The movie suffered from "Director So Famous That Editors Were Too Afraid To Tell Him To Cut It Down Syndrome." I'm sure you've seen other movies suffering from the same before. Eyes Wide Shut, for example.

Alexander would have been a great 110-minute movie.

Re:Ha ha

[cloudmaster]

Maybe that firm could devise a service wherin they rent out processor time on their grid during that idle time! It'd be great - people could just use what they need rather than investing in a complete grid that they wouldn't totally utilize themselves! Wait...

Price too high?

[GGardner]

There was a lot of debate the last several times this was posted about Sun's $1/cpu-hour price, how TCO is a lot more than hardware cost, etc. Still, a google search reveals a bunch of other companies who lease out CPU farms (mainly intended for rendering), who charge less than $1/cpu-hour.

Re:Price too high?

[Frank T. Lofaro Jr.]

Charges by cpu hour is a bogus metric.

$1 a CPU hour on a 486 33MHz is NOT a better deal than $5 for an hour on a P4 running at 3.2 GHz

It might be good for billing but it is bad for comparisons.

Re:Price too high?

[Mr. Sketch]

Just curious, what search terms did you use? I tried googling for something similar a while back and couldn't find very many companies that did this or would advertise their rates on their webpage.

Re:Price too high?

[GGardner]

I have no affiliation with any of these companies, but try something like "render farm price Ghz". I see prices like 30 cents per ghz-cpu-hour, in volume, including use of rendering software (dunno if the software is open source, or not). 30 seconds of searching found three companies, so I assume there's a bunch more.

Re:Price too high?

[southpolesammy]

For those of you too lazy to do the math, this comes out to $8,760/cpu/yr, or for a typical 2-way server, $17,520/yr. Over a typical 3-year lifecycle (YMMV), this is $52,560 in expense for a 2-CPU server. Of course, this includes administration of the service, such as backups, sysadmin, power, data center space, etc...

Compare this to buying a 2-way Sun V240 at about $7,245 (pre-discount), and you have $45,315 worth of TCO cost-savings to justify to management over the same 3-year window to make this worthwhile. Now I don't pretend to speak for others, but our SA's administer multiple systems, typically at least 20/SA, so unless your SA's make more than $300k/yr, I can't see this being feasible.

Re:Price too high?

[Surt]

Looking at it on a per cpu per year basis doesn't capture when this would be useful. Suppose you needed 1000 cpus for a week, but then your data analysis project would be done for the year. Better to buy yourself the 1000 cpus or to rent them?

Re:Price too high?

[hackstraw]

There was a lot of debate the last several times this was posted about Sun's $1/cpu-hour price, how TCO is a lot more than hardware cost, etc. Still, a google search reveals a bunch of other companies who lease out CPU farms (mainly intended for rendering), who charge less than $1/cpu-hour.

You've also got to consider that Suns are slow, so its even more expensive. I'm not just saying this, I've run tests on everything from an UltraSparc2 @ 333Mhz to an UltraSPARC3 @ 1.2GHz. They are particularly slow in memory bandwidth. They are not as well supported with 3rd party hardware (eg, interconnects) or software as Linux is. Also, their newer processors do not come with as much onboard cache as they used to. I've seen some programs run faster on our slowest machines because of cache coherency. This is a little different now that Sun has gotten on the AMD bandwagon, but those are pretty much commodity that can come from anywhere, and all in all Linux is generally better than Solaris at HPC kind of applications. Solaris is a better operating system in terms of maturity and whatnot, but I and most others agree that Linux is a better choice.

Re:Price too high?

[petermgreen]

yeah the price is probablly a bit on the high side. though you didn't account for the fact that your own machines are unlikely to be at full load doing usefull work 24/7.

how much do places like specialist render farms charge by comparison

Re:Price too high?

[LurkerXXX]

As others have said, this service isn't targeted at folks who use their CPUs 24/7. Those folks already have their own clusters. This is for folks who occassionally need massive computing. Your arguing numbers for entirely the wrong market. Know your market.

Re:Price too high?

[slavemowgli]

One case where it would be feasible is when you only need very little computing power. An admin may not cost much per machine when you have a lot of machines; but when you only need one or two, you can't hire 1/10th of an admin to deal with them, you still have to get at least one. So in that case, Sun's offer would make sense.

Re:Price too high?

[Alioth]

That comparison is only valid if you're going to keep the CPU on your Sun V240 pegged at 100% for a year. Realistically, if your V240 is depreciated over a period of 3 years (quite common for computers) you'll need to keep the CPUs pegged at 100% for at least 6 months of these 3 years to get a better TCO (on just the hardware).

Especially if you need the power of a 300-machine cluster - then figure out the cost of the HVAC, depreciation of the HVAC, cost of the building to house it...the list goes on. If you just need a big cluster for a few weeks of processing, it really doesn't make sense to do it yourself.

Re:Price too high?

[Decker-Mage]

Actually that isn't true at all anymore. One of the things I do here is act as that 1/10th of an admin and price accordingly when I bother to charge at all (I don't charge non-profits for instance). That's why I developed my techniques for securing and remotely monitoring systems here. Actually mostly it's just monitoring these days and I'm not the only person/firm doing this in the IT arena. SMB's and consumers can't afford a full-time IT person so the market is out there. Sun's premise is that no one else can provide an equal level of service and support. In that they are sadly mistaken and on the per client basis, with client hardware, I can beat them badly. There is a niche for the Sun product, unfortunately IBM is already playing there and they have clients lined up like crazy.

New system, new market, enterprise products

[jiushao]

Not that suprising at this point, it is a new system, it tries to create a new market and it does so in the enterprise space. Things don't suddenly catch on when it comes to enterprise data service, Sun has to offer the service to get the talk going and in another few years we will really know if it turns out good or not. It is much too early to make any judgements about the feasability of the project.

Personally I think the idea might work, but it might not in this incarnation. There seems to be a fair chance that Sun can claim to be ahead of its time again, which has in some ways been a while. Which is a good thing in itself, Sun has historically been a nice company to work with but has suffered from some stagnation for a number of years.

Re:New system, new market, enterprise products

[Red Flayer]

"Not that suprising at this point, it is a new system, it tries to create a new market and it does so in the enterprise space. Things don't suddenly catch on when it comes to enterprise data service, Sun has to offer the service to get the talk going and in another few years we will really know if it turns out good or not."

Exactly. It's not like within a company, the demand for ridiculous amounts of CPU cycles materializes overnight. Suddenly, Production Dept says to CIO, "Hey, we need to render two hours of animated video in three days two months from now! How do we do that?"

This'll become more popular when current render farms and computing clusters become obsolete, and not likely before then.

I'm sure there are governments that could make use of this service, and I'm also pretty sure that they would be able to get Sun to not publish their use of the grid.

Re:New system, new market, enterprise products

[Minna Kirai]

I'm also pretty sure that they would be able to get Sun to not publish their use of the grid.

If you're so paranoid that you don't want enemies to even know you used a 3rd party for your compute jobs, then you're already far past the point where you'd trust a 3rd party to handle the job.

Re:New system, new market, enterprise products

[Red Flayer]

But how do you hide your purchase of a server farm?

Most of us have friends and family

[Work Account]

Many intelligent developers like Bram Cohen, the creator of Bittorrent, didn't have much while they were developing.

IIRC he spent a year or two living frugaly with relatives or friends because he knew he had a great idea and wanted it done as soon as possible.

Sure, he could have used some money, but he wasn't about to get a job and then have the company own his $8.7 million dollar idea (and that's just the current market value not including future potential revenue streams).

Re:Most of us have friends and family

[RingDev]

So the answer is for coders to live off well-fair and with their parents?

Personally, I think that idea sucks. I'd much rather atleast get some VC to survive off of and launch an LLC to get the product up.

-Rick

Re:Most of us have friends and family

[Guspaz]

No, he did so because he focuses on a problem to solve at the expense of other things. He didn't intend to turn BitTorrent into a business, and only founded the company because his father kept bugging him about it. He has Asbergers, a form of Autism, and it's an obsession with solving a problem that leads him to do what he does, not business sense.

He didn't sit down and said "Hey, I have this great idea for content distribution that I think I can make money from."

He's said this numerous times in various interviews.

Re:Most of us have friends and family

[Shanep]

He has Asbergers, a form of Autism, and it's an obsession with solving a problem that leads him to do what he does, not business sense.

Ah, last I heard, he had self diagnosed himself, thinking that he had "Aspergers".

Have his thoughts on this been confirmed by a professional?

Re:Most of us have friends and family

[slavemowgli]

You mean "he thinks he has Asperger's "(with a p and an apostrophe - the guy was named Asperger). He's never actually had it diagnosed by a doctor, so even though it seems likely that he's indeed suffering from it, it's not actually confirmed.

Re:Most of us have friends and family

[birge]

If he's really got it, he's the highest functioning autistic that I've ever seen. I know a person with Aspergers, and it was hard for them to get through school. Furthermore, their obsessions tend to deal with obtaining weather information, compulsively, for every city in the US every day, as opposed to sitting down over a period of years and conceiving and writing a very complicated transfer protocol. My guess is he's a very talented guy who also happens to be a bit of a headcase. But who the fuck self-diagnoses with autism? It's almost an oxymoron. If you've got it, you're probably not aware you've got it.

Re:Most of us have friends and family

[Achromatic1978]

he had self diagnosed himself

As opposed to self-diagnosing someone else, you mean?

Re:Most of us have friends and family

[triso]

You mean "he thinks he has Asperger's....He's never actually had it diagnosed by a doctor, so even though it seems likely that he's indeed suffering from it, it's not actually confirmed.

Are you casting Asperger's on my diagnosis? With apologies to Curly Howard.

Re:Most of us have friends and family

[DeathElk]

Mmmm, ass burgers

Re:Most of us have friends and family

[icantsurf]

Ah. . . to live in the nineties again. What I wouldn't give. Quite frankly Venture Capital isn't entirely reliable or as adventurous as it was 10 years ago. Living with family or on government assistance (those of you who got laid off and went back to school to take out student loans qualify here) became a reality for a significant quantity of the workforce.

The trick is to develop smaller projects to pay for ramen and the bills while your brainchild works it's way to beta. Eventually you will either make millions on your IPO or release an antiquated Rube Goldberg. Either way you don't owe nothin' to no body. Meanwhile I'm looking for a cardboard box that houses 3. Anyone purchased a Fridge lately?

Re:Most of us have friends and family

[Guspaz]

I went to high school with somebody who has Asperger's, he was obsessed with gaming and it was almost impossible to get him to stop. Of course that was just one of the things. He had other obsessions. At the lunch table if you took his lunch box and moved it to the other side of the table he'd go crazy trying to get it back; like he had lost all self control and his only goal in life was to get that lunchbox back. Even if it was empty.

Not everybody with the same condition is exactly the same.

Re:Most of us have friends and family

[birge]

I agree everybody manifests differently. But I think one thing is fairly common: people with Asperger's autism generally don't seem to be nearly as high functioning as Bram clearly is.

Re:Most of us have friends and family

[Guspaz]

Really? The guy I knew kicked my ass at Age of Empires 2 every time. He seemed pretty high-functioning. I quote the wikipedia article: In pursuit of these interests, the person with Asperger's often manifests extremely sophisticated reason, an almost obsessive focus, and eidetic memory. Hans Asperger called his young patients "little professors" because he thought his thirteen-year-old patients had as comprehensive and nuanced an understanding of their field of interest as university professors. One thing to keep in mind is that Asperger's is one of the mildest forms of autism, and doesn't mean people can't be high-functioning autistics. Many are.

On the other hand, the Wikipedia article claims as a fact that a bunch of people have Asperger's (like Bill Gates) when in fact there exists only speculation that these people MIGHT have Asperger's.

Anyhow, I must admit that I've been somewhat skeptical about Bram myself. Self diagnosis is a tricky thing. But from descriptions in interviews of how he's behaved, he has SOME sort of condition.

Important Question

[Tiberius_Fel]

An important question is whether this is a failure of marketing or a poor choice of target market. If the target market exists but is not using it, then you might be able to consider it a failure of marketing: There is demand and supply, but the demand is not aware of the supply. If the target market does not exist, then Sun has obviously chosen to go into an area which is not a worthwhile venture, at least at the present time.

Though, it's possible that the target market hasn't been formed yet and Sun is going for the "If you build it, they will come"; i.e. by creating the possibility they will generate demand for it in the future.

Re:Important Question

[LnxAddct]

No, the problem here is price. Its $1/per hour/per cpu. If I have something that takes on average 60 days to calculate using a 20 node cluster and I do this yearly, it will cost me $28,800 per year to use Sun's grid. I could buy a 25 to 30 node grid for that kind of money with comparable node to node performance. So I'd be getting more nodes to work with and I would have a cluster that can be used for probably the next 5 years of that same computation, and during the times where its not needed for that computation it could be used in other areas. Also, it could be in a nice little rack or two in the corner of the office somewhere, so it taking up space is not an issue. The cost savings, plus benefits, of keeping it in house is astounding, just from raw hardware costs you save $115,200. Both clusters would have to be administered, but you have a much higher level of control in house, not to mention porting is probably easier and you can be certain that you won't have to port again for at least 5 years. In order for Sun to make this cost effective for anyone, they need to make it outrageosly cheap, like 1 or 2 cents per hour per cpu, and that still might not be enough to overlook the advantages of having an in house cluster.
Regards,
Steve

Re:Important Question

[Lucractius]

when factoring the cost did you factor in the cost of power, both to run the machines and extra air con load, along with the expected/allowed maintenance costs? its cheaper yes... but not that bad.

and while i agree it coudld certainyl stand to be cheaper, perhaps half its current price, its about total value not about raw power.

What i want to see is a way i can pay 5 bucks to get someone ELSE to do my distcc builds for me while i spend 5 hours gaming instead of watching my machine build at high cpu load and leave me some time to relax. or even i pay 5 bucks and i get it done in 1 hour instead of 5 ... see the savings and usefulness there.

Re:Important Question

[hackstraw]

An important question is whether this is a failure of marketing or a poor choice of target market.

This is not a marketing issue. For people that need cycles, they will go wherever they can to get their needs met. They will come in record time if there is no wait in the queue. The price seems a bit high to me ($876,000 for 1 year of 100 CPUs which is not that many), Suns are a little slow compared to other machines. Most people use Linux for these kinds of applications, so there may be porting involved.

If you don't know researchers that need computer cycles, you don't know how much of a desire they have to run as much as they can wherever and whenever they can.

Why use Sun when Zombies are cheaper?

[G4from128k]

I bet that not that many people actually need extra CPU cycles that don't want those cycles on a more permanent basis. Perhaps the only people that need short-term access to computer power are fly-by-night spammers and DDoS extortionists.

Grid computing is cheap really

[linzeal]

Not exactly, but one of many reasons methinks. Another one is that people have a real problem running anything with their numbers on someone else's hardware period. If they are thinking that this is going to fly in the US where a great majority of adults drive their own cars to work even in cities with public transportation, own 3400 square foot McMansion instead of a 2000 square foot Victorian or in general prefer their own toys to play with they are smoking better stuff than me and I'm in Humboldt. Sun should of lent the platform to be sold as a software solution that would run on existing hardware like the office computers at night. I'm currently writing up a white paper for a grid computing solution at my new campus and it has nothing to do with purchasing any hardware or software and everything about using the computers we already have more effectively.

Re:Grid computing is cheap really

[petermgreen]

i'd think on a campus you'd have a lot of computers that with appropriate software setup could be repurposed at night. (you have roomfulls of pcs for students to use don't you? are ALL those rooms open 24/7 i doubt it so just set it up to reboot after closing time into your clustering install and either shut down or reboot back into the standard install afterwards).

Anyone notice it's still in Beta?

[quadra23]

In fact, however, Sun is still in beta with this CPU program and not set to launch a publicly available utility computing system for weeks...It seems hard to believe that Sun would pass on the opportunity to dangle such a user in front of the press if it existed given that we're 14 months away from the utility computing launch date...The company promises us such a day is coming sooner than later and that it will have plenty of customers to name in the near future. Still, given that it took a year to push the program to a beta, one wonders how long an actual living, breathing utility center will take.

I also must agree with anyone here who mentioned that you need to pick up customers during the development of the project and not just assume they'll come from no where. If you have at least one customer then they'll do far better publicity for you then thousands of dollars worth of marketing ever will. Sun might have a big name but that doesn't immediately get people jumping to their project. You don't dictate to people what they *should* do you make them think that using your product was *their* idea and that your just there to fulfill the customers requirements.

Re:Anyone notice it's still in Beta?

[rathehun]

well - they are in partnership with Google now ;)

ChickenEggWare

[Doc Ruby]

They should make it free to attract developers, then hype the service, grandfathering in people when it starts to take off. As it is, no one has apps, or even app ideas, ready. Sun's marketing people should know that, but failed. They have a second chance to do it right.

Not suprising, just do the math...

[nweaver]

If you have tasks that can be done on compute farms, computer farms and clusters have gotten relatively easy to manage and deploy and are CHEAP.

Sun's charge of what, $1/CPU-hour is just way way way out of line compared with what you can build yourself (using dual core, dual processor athlons from Sun, for example), if you have any consistant demand.

Re:Not suprising, just do the math...

[moviepig.com]

Sun's charge of what, $1/CPU-hour is just way way way out of line...

...and completely neglects time-honored marketing principles. Should've been $0.98/hour.

Re:Not suprising, just do the math...

[john83]

It does sound like a lot to me. I imagine the kind of people they're aiming at are researchers with a grant to perform some research, but who wouldn't need (and couldn't afford) as large a cluster as they need. The article from El Reg says that they are getting some private jobs from companies. "The reason you haven't heard about any Sun utility customers is because most of them have avoided the company's publicity requirement by avoiding the original product. A number of companies in the financial services and oil and gas sectors have purchased large quantities of Opteron CPUs from Sun, MacRunnels said. These customers negotiate their own price for the processors, tend to use the chips all the time instead of popping on and off the grid and refuse to reveal their names to the public. Sun has started to call these "commercial" utility computing customers. The company assures us that some of these commercial customers do pay for these utility computing services. They use huge blocks of processors to crank through models such as Monte Carlo simulations. Sun won't name any of the customers or say how many it has other than to declare "the figure is in the tens" of customers."

Re:Not suprising, just do the math...

[nweaver]

And these commercial companies are probably paying a LOT less:

They are effectively leasing a cluster computer (leasing has tax advantages), and probably their rates are much closer to the real value:

EG, a 2core 2cpu SunFire X4100 is $7500. Lease probably amounts to ~1/3rd of that plus a little extra (lets say $800/cpu-year). Power for the beast is ~600W, so 150W/cpu. At $.20/KWh (includes cooling and reliability), thats ~$250/cpu-year. Lets add $100/cpu-year for the physical location and maintinence, that totals out to $1150/cpu-year.

So Sun probably charges these dedicated customers more like $1150-$1500/cpu-year (perhaps a little more than what they'd pay to build the cluster themselves, but they save hastle factor and having to find space). But this is VASTLY less than sun's retail price charge of $8760/cpu-year.

Re:Not suprising, just do the math...

[dommer2029]

I'm confused about your math.

Ah. I see now. You're saying that the $7500 box is $1625/CPU.

So the retail price of buying and running 1 CPU is $1625 + $250 + 100 = $1975. Which is still not $8760.

$7500 + 4 * $250 + 4 * $100 = $8760? Nope, that's $8900. Not sure what you're sayin' here.

Anyway, yeah: $1/CPU-hour is nuts. It's so far out of line with the cost of purchasing that you would always want to buy with your budget, take twice as long to run your first job, and then re-sell your hardware or keep it around for the next job.

Pssshhh thats because

[caffeinex36]

"My Zombie network has 1000000000x the computing power of suns grid....

and you can use mine for some good CC numbers. any company CLEARING doing a cost benefit analysis realizes that its much cheaper to go with me."

-Founder of P0wnd Zombi3 N3twerkz

Re:Cost of porting, uncertain future of the servic

[immovable_object]

I totally agree. No effort has been made to figure out the security aspects of the solution. How do I know that my data is distinct from any other companies' data? Also, when it comes to big compute farms, doesn't that mean that I'm working on HUGE datasets? How do I get TB's of data to the gride farm and then get it *back*?

Seems like a "you build it and they will come" mentality. In the days of laptops with good compute facilities, I have to think this represents dinosaur thinking.

Doesn't Jiva do this cheaper and faster

[abbyhoffman]

I've looked into grid computing a few times and ran a few clients as well. It seems that Jiva does the same exact thing, but much cheaper. Then again there is also Parabon and united devices, though they tend to charge even more than Sun.

Re:Not a question of price, but privacy, latency

[xtracto]

Why on earth would any company ship their valuable data to a third party to process?

Completley agree with you, moreover the companies (as stated on the article) that could use this service are finance focused companies and maybe some phramacy companies.

This makes me think on the software laiability issues point, I am sure these companies would demand something very very far from the typical "EULA" or contract to use this service.

Darn I am sure any of the big stock exchange palyers would be really pissed of if someone was sniffing their data during the transit from Sun clusters to their clients.

I think a good way for Sun to make this service go up is that a company could rent them this CPU power and lease it to smaller users. Something on the lines of an Eceed UNIX client whose "virtual" servers run on these Sun clusters.

I remember I saw a talk on this system on the TADA0-IJCAI05 workshop. IIRC they are planning to give away CPU!

Re:Cost of porting, uncertain future of the servic

[$RANDOMLUSER]

Nah. This was for short-term high-CPU demand projects, like genome sequencing and protein folding.

doesn't even let me request and account

[menix]

I'm on OpenBSD with KDE and when I click on the button to register for an account, nothing happens. There's the reason, bad programming. They don't have customers because people can't register! JERKS

Re:doesn't even let me request and account

[elmegil]

Did you pay attention? It's in BETA. Sun's not doing an open "Google-style" BETA, they're restricting it. Whether that's a good choice or not, it's pretty obvious and not worth whining about don't you think?

Perhaps this is just a "hey its there, sell it!"

[aphaenogaster]

I am sure Sun had massive computer power just laying around, they probably realized they could leverage it as a commodity while investing nothing but RandD they were going to do anyway. If MS and Google and all plan on huge ajax like projects, sun may very well have something in the future. :"who needs more than 64kb anyway!"

Problem is...

[dshannon]

... that I (as an EA) don't really understand the proposition and what I can do with it. Sure I've read the blurb, I've even been to Menlo Park and had the presentation, but the question I want to answer is *what* of all my core apps I'm going to run on it. Do I get to go to Oracle grid on this stuff? Can I run all my core back office apps on it? What do I pay on top of the $1/cpu/hr? Bandwidth back to head office?

On top of all that, it's clear that I'm not going to abandon our existing investment in Sun hardware to take immediate advantage of this while that hardware still has a leasing life of 2-3 years. Sure I'm interested, it doesn't particularly benefit the company to have a stack of office space devoted to a computer room, and it's harder still when the business grows fast and we constantly need more gear. But Sun aren't in my face about this stuff, aren't giving me the numbers I need to take it to the CIO. When they do, then I'll think about it.

On the other hand, Sun are to be congratulated on their other initiatives in this kind of pricing model. To an enterprise with small numbers of staff but high revenue, their per FTE/yr software licensing on Java Enterprise System et al is a wonderful model which many other vendors will have to catch up with as we move to multi-core CPU's as standard. For us, the other J2EE vendors just can't compete on price (FOSS excluded of course).

Utility computing is coming, let's face it - but mainly it's a question of education of the masses, and time to get through hardware replacement cycles. Of course I'm a bit surprised that there's NO customers yet, but that still doesn't mean there won't be, ever.

Re:Problem is...

[Decker-Mage]

Nicely put. My problem with this model run along the same lines. I can't see either the engineering or economic justification yet although I'm sure there are some if Sun would give us their rationale. If the rationale is simply that IBM is doing it, and doing it rather well from my data, then they have a real problem. If they have identified some niches in the IT economic pie, could they please explain what they are? Pretty please? I have a lot of people around the world that ask me for analyses on issues like these and without data I can't recommend it. However this isn't a problem unique to Sun by a long shot. As Number Five said: "Need Input!"

Why not use BOINC?

[Bob_Villa]

Couldn't a company work something out with BOINC where they pay BOINC $.01 per CPU-hour, and $.01 per milestone to each participating member?

I bet people would sign up in droves if they could earn a little money for their free computer cycles. It could be paid quarterly or monthly using an online payment service like PayPal or through good old fashioned checks in the mail.

Just an idea, and for only $.02 per hour instead of $1.00 per hour.

Dave

Re:Why not use BOINC?

[Bob_Villa]

But is it a loss if you were going to have your computer on all of the time anyway? Or if you are using your computer and your spare cycles go towards crunching their project? I turn mine off at night, but some people leave their computers on 24/7. Or what if it was more for each milestone.

It was just an idea to think about. ;)

Re:Why not use BOINC?

[Wesley Felter]

Try running an MPI app on BOINC and let us know how it goes. Or an app that requires 4GB RAM and 10GB of disk space on each node.

It's all about the applications.

[khasim]

People don't buy a computer to run a processor.

People want to run applications.

The thing I'm not seeing in Sun's model is anything about the applications. Are they off-the-shelf? Who installs them? Who maintains them? What OS's are available? What security is available? How can I make sure that no one else sees my data?

We've already been through with with the Application Service Providers (ASP's) and there are still a few out there making money by providing Internet access to their apps, running on their servers, storing and processing your data. Payroll is an easy app for that.

I think Sun is missing part of the equation.

Re:It's all about the applications.

[sr180]

I think that you have missed the target market for this grid. It isnt about the applications at all. This is about having a complex problem that requires a lot of computation power for a solution. Folding at home is a prime example. Say that you are a drug company and need to investigate the folding patterns of some new drug. This is going to take 87600 hours of processor time (10 years on a single processor.) This is only a one off requirement so its not worth buying a grid network yourself. Using 10,000 processors on the sun network, you are only going to have to pay for 10 hours or so. (providing the problem can be sufficiently parallelised.)
You dont give a stuff about applications. You arent providing them, you are simply buying raw computational power 'ON DEMAND'

No default applications. No updates for them. Just an OS and a way of running your code to start crunching the data you want to crunch.

Sun is not looking at the part of the equation you are looking.

It's just "time sharing", and it's obsolete

[Animats]

There used to be a scientific time sharing industry, with mainframe computer time rented by the minute. It's dead. Most commercial jobs you can do on PCs. If you have an ongoing need for more crunch power than that, you can get your own computing power, and it will be cheaper than renting it. The market for huge numbers of intermittent cycles is weak to nonexistent. The basic problem is that there just aren't many companies with giant number-crunching jobs for which they are willing to pay. For the same reason, there are very few privately owned supercomputers. There was a "grid computing" utility about two years ago, before Sun tried it, and they didn't get customers either.

Sun's "grid computing" operation seems to be an attempt to find a use for unsold Sun servers, or at least to avoid writing their value down to scrap prices.

f you went to a big hosting company and said you wanted a thousand unlimited-CPU-at-low-priority shared hosting accounts, valid only from 2300 to 0700, you could probably get a really good price. If "grid computing" were useful, somebody would be doing this. All those nearly idle CPUs could be doing something.

There's a successful grid computing company: Akamai. What they sell is distributed hosting and cacheing, which they call "Akamai On Demand Managed Services". When the web site for the World Cup or NASCAR or Britney is getting millions of hits per hour during some special event, thousands of Akamai servers switch to serving those pages to handle the transient load. That's a successful "grid" application, and it's been working for years.

Akamai does more than serve pages. You can run your business logic, in Java, on their servers. So they're already set up to run user code on their grid. If anybody is going to sell grid computing profitably, it's Akamai. They're all set up to do it. Yet they don't.

Re:It's just "time sharing", and it's obsolete

[ChaosDiscord]

The market for huge numbers of intermittent cycles is weak to nonexistent. The basic problem is that there just aren't many companies with giant number-crunching jobs for which they are willing to pay.

It's a small market, but not nonexistant. Anyone doing high-energy physics needs as much processing power as they can get. Companies doing genetics research (say, researching gene therapy) tend to need lots of compute time doing massive searches and comparisions of genetic databases. Insurance companies doing simulations and analysis need massive computing power. Special effects companies chew through computer time.

There is no question that massive amounts of compute power are needed. The question is: is it actually cheaper to rent the CPU time instead of just buying and managing the machines themselves? I'm less certain on that. While someone else has to worry about buying and maintaining the machines, you need to modify your workload to work on machines you don't control. The remote site may upgrade to an incompatible system to serve other customers. They could configure themselves to run whatever OS loadout you want, but that will cost more to setup and maintain. You typically need to send your workload across the public internet; putting gigabit ethernet between your cluster nodes so you can toss 2 gigabyte data sets around is relatievly ship. Getting a big enough network connection to set those datasets across the country is more expensive. Running over the internet is also more fragile. Oops, a backhoe just took out the connection. When something goes wrong, why does the provider care? Doing it in house means you have local staff you can lean on. A provider can be made to care, to provide guaranteed response times, but it'll cost you even more.

Re:It's just "time sharing", and it's obsolete

[tacolicker]

It's a sad day when our most popular events involve NASCAR hicks and Britney.

Re:It's just "time sharing", and it's obsolete

[Lucractius]

Funny thing is knowing Sun and solaris 10 i wouldnt be surrprised if they happily let you take your pick, running the grid to be maximum flexibility, providing back/forwards solaris compatibility, BSD/UNIX compat, Linux compat, and probably even Mac and Windows Compat somehow possibly on other servers or wine or soemthing... who knows... they do i dont but i know they sure arent stupid when it comes to making things that work.

Re:It's just "time sharing", and it's obsolete

[Fnkmaster]

You missed the key word: "intermittent". The point is that most of the firms and people you cite need that kind of computing power regularly, not intermittently. If they need it more than 4 or 5 weeks a year, it makes more sense to buy than to rent from Sun. And several of the applications you describe already have niche service providers offering applications experience AND rentable computing power, which provides customers with a lot more value than Sun is.

Re:It's just "time sharing", and it's obsolete

[ShakaUVM]

> It's a small market, but not nonexistant. ...

>Companies doing genetics research (say, researching gene therapy) tend to need
> lots of compute time doing massive searches and comparisions of genetic
> databases.

Except that's exactly the problem. You have this massive database. You can't just wave a magic wand and have the database appear on Sun's Grid. You have to upload it to them. Which takes, what, weeks? Years? Sure, maybe then their computation will be 10x faster, but you've lost the game from the start.

The performance comparision for the Sun Grid is:
Time to Upload Data + Time To Compute + Time to Download Data

vs

Local Time to Compute (since the data is already on your servers, presumably)

When you look at it this way, the Sun Grid becomes a lot less attractive.

Real grids solve this in various way, with data routing algorithms controlling where data is pushed out to -- but it works because your local computers are on your own grid with a high speed interconnect to the other players. Sun isn't part of your own Grid, so you can't employ these data routing strategies. And presumably Sun won't accept crates of DVDs in the mail and manually upload them to their server farm.

They're also in competition with the (free) academic grids.

Re:It's just "time sharing", and it's obsolete

[Minna Kirai]

It's a small market, but not nonexistant. Anyone doing high-energy physics needs as much processing power as they can get.

"As much as they can get" is very different from "huge numbers of intermittent cycles".

Academic phycists (like your link) aren't going to suddenly notice that they need 20,000 Ghz-hours within two weeks. If they had that kind of spurt demand, they'd be a good customer for Sun's grid.

Instead, users like that will have a fixed annual budget, and will try to maximize their Ghz-hours. Currently, you can get more computation per dollar by purchasing hardware, instead of leasing from Sun. That is unlikely to ever change, because Sun's pricing has to include overhead for the flexiblity of the grid, big bandwidth to move the data in and out, and liability/security costs to assure customers of their privacy.

Only if Sun achieves a superb economy-of-scale with administration will they ever be more affordable than in-house servers or other, simpler CPU rentals.

PS. Academics have a subsidized supply of white-collar labor in the form of grad-students, who can adequately admin the bought hardware for little cost.

Re:It's just "time sharing", and it's obsolete

[dkf]

The academic grids aren't free. It's just that it's the Principal Investigators who get the bill, and not their jumpsuited minions, err, research students.

Re:It's just "time sharing", and it's obsolete

[ShakaUVM]

They are free in the sense that they are looking for "customers" half the time so they can share with the world how awesome their 'supercomputer' is. It's how they get more funding. "We cured 2 cancers with a 1000TFlop grid, so we could cure 10x more with a 10x more expensive one." ;)

So when Sun comes a-knocking, it's not nearly so appealing since it's not, you know, free and all.

Re:It's just "time sharing", and it's obsolete

[Biolo]

There is absolutely no way that Sun are doing this to try and recycle 'scrap' machines. I happen to know what the machines they are using are, and they are fairly newly released ones that are a range that is manufactured by OEMs, not Sun themselves. These machines are being manufactured to Sun order, not coming out of over-production. They are also putting a massive investment into the infrastructure (datacenters, racks, etc). I wish I could tell you more, including the machine model involved, but I can't (morally, promised I wouldn't, no legal restriction). I can tell you that they are putting in just one model. If they were trying to recycle scrap material there would be a huge mix of machine types going in, not just the one.

The one thing I do wonder is why they aren't joining all of their internal systems into the grid and selling that unused capacity. I know they are making that sort of a grid, but it's internal research use only AFAIK, and only a relatively small number of systems.

Smartest Reply I have read so far

[kninja]

There were many "The engineers did it because it was cool and no one would buy it" comments. This might be the case, but Sun does have a marketing dept - and I'm sure that at least someone knew about it and did their homework.

That said, it is obvious someone in the marketing dept didn't get it right, but at some point it comes down to luck.

They are trying a new product in a new market space and it might fail - due to any number of reasons. The two that come to mind are: The customers don't know they need it yet (as parent said - ahead of their time), or that a competitor does it cheaper or better. This happens all the time to small startups, divisions within companies get axed, etc.

"More companies fail from a lack of customers, rather than a lack of product"
(I'm not saying Sun is going to fail, but this division might get axed.)

Welcome to the business world, it's vicious, yet rewarding if you do it right ('Google', 'Ipod' division).

Not for science

[PineGreen]

I know why they don't get any science customers from my own experience. Basically, if you buy a cluster of your grant, you pay just for the hardware, everything else: electricity, cooling, network, comp support comes from the department's budget. These costs are not negligible.

If you tried to buy time from Sun, then everything goes from your budget... So, for an average scientist, who might be interested it is much cheaper to buy my own little cluster and piggyback on department's infrastructure...

Re:Not for science

[hde226868]

This depends really on where you're working. In many universities the cost of operating the computers COMES out of the budget of the scientists. It is called the "overhead", which is normally between 50-80% of the total value of a typical research grant in the US.

In the UK, where I work, the research councils have just switched over to "full economic costing", and again often the biggest item in research applications is due to FEC. For example, for a typical application for a postdoctoral position, the annual salary will be on the order of 20-25kGBP, while the FEC for office space, secretarial and other janitorial support, computing etc., is about the same per year. So, no, when applying for research grants, we do have to take this into account, and it is usually the most dreaded part of a research application and takes up similar amounts of time than writing the scientific justification of a proposal.

Now, if you find a way to hire, e.g., a graduate student, operating the computers might come cheaper than a full blown sysadmin, but even a graduate student will cost you about 25-30k$ per year in the US if you figure in the tuition fees.

So, from this point of view, one can argue that Sun's offer is still expensive, but the operating costs of a research group are not negligible and I think that overall Sun might only be a factor 3 or so more expensive than operating your own computing center.

The real reason why scientists won't jump on Sun's offer is that we have the right to get computing time for free from the national supercomputing centers. If you have a computing intensive project, getting time there is not too difficult, so there is no need to go to commercial providers.

Re:Not for science

[hackstraw]

Basically, if you buy a cluster of your grant...

That is also key. A grant is usually very specific in terms of buying equipment and paying people. Paying for a service is something I've never heard of being allowed on a grant.

Re:Not for science

[John Newman]

I think another reason is that computational labs have their own hardware (or shared core facilties) while non-computational labs with the occasional need for number-crunching find it easier to do it on their own general-purpose hardware (slowly) or borrow time on a friend/collaborator's hardware. If you're running so few jobs to make it not worth your while to buy dedicated hardware, it seems unlikely that you'll care if a job takes a few days on your lab workstation, vs. a few minutes/hours on Sun's Grid. Plus the fact that labs which can't or won't buy dedicated hardware are likely to be small and narrowly-focused, and lack the expertise to quickly adapt their code to run on the Grid.

I suppose the only exception would be HPC labs - with the need for massive computing resources and the expertise to adapt to them - without any access to academic HPC facilities. But I can't imagine there are many of those out there. I mean, what would they have been doing up until now?

Sorta gives credence the idea that ownership...

[phorest]

This gives me the idea that all the talk about hosted apps/services/storage is way overblown. Google shows their hand M$ bites, nothing to show for it (yet); yet let's create some more buzz how you should keep your stuff in some far-away place with little control.

I believe that even though it sounds good, people are reluctant only because ownership physically changes hands. Which in the business world that is an important distinction. I know this may seem offtopic but this posting really does point in this direction.

Maybe with the history of easily being able to access your servers/workstations via a multitude of ways (term servs, RDC, VPN, Citrix... etc.), the market just isn't getting worked up about it. Which in M$'s case, you get that ability standard with as little as 1 online XP box (RDC), but for us with servers and domain controllers RDC is a godsend...

Me, as a business owner will gladly deal with licensing fees for a variety of uses instead of leasing space for a monthly/subscription price. Even though EULA-wise you technically don't own anything M$, but physical possesion alludes that it is a tangible asset.

Re:Not a question of price, but privacy, latency

[steve_l]

You give it away to somebody you trust, who locks down the cluster and gives you a private datalink, in exchange for CPU power you cannot afford yourself. And those big datacentres ought to have big private filestores for you to use too.

I dont know about pay-for-grid, but we host CERN-originated Large Hadron Collider simulations on our cluster, as do other sites all over Europe. There isnt the heed to build a single giant linux cluster, when you can use spare cycles from high performance across the entire continent.

-steve

Budget Accountability

[mjspinks]

SUN seems to have missed the mark. Even in research, budgets are an issue. If an organization has n dollars to spend in solving a problem, results are expected using n dollars. When you purchase equipment, you can re-compile /re-run the job without adding to your expenses. In order to properly budget a research task (in terms of CPU time) you must know the amount of time required to solve the problem. If you are responsible for the decision to use the SUN Grid and your problem takes longer to run than expected (even at $1/CPU/hour) you could get intro trouble.

For example:
A problem that runs 25% longer than expected
Budgeted: 10,000 CPUS x 16 hours = $160,000
Actual: 10,000 CPUS x 20 hours = $200,000

This 4-hour / $40,000 overrun could prove detrimental to one's career.

Target Market is the problem

[puppetluva]

Anyone who is savvy enough to need GRID computing is savvy enough to build their own grid very cheaply. Sun's GRID would only be useful for times where one's own grid is overloaded for brief periods of time and you don't want to scale up (a confluence of factors that is very hard to predict and order from Sun ahead of time).

I'm surprised that there wasn't more of a business analysis of this ahead of time before they plunked down a ton of money to make it happen.

Re:Target Market is the problem

[Shanep]

Anyone who is savvy enough to need GRID computing is savvy enough to build their own grid very cheaply.

Yes, there are lots of desktops out there in corporate land. A lot of CPU's which are mostly idle, even during the busiest parts of the day. Word and Email don't chew up many cycles, although I imagine surfing the net could chew up lots more due to flash, etc. Then there's after hours...

lots of reasons

[bcrowell]

Lots of possible reasons for this:

1. Grid computing is a different style of computing. It requires a lot of software work to set it up, and not many people have experience with that style of programming. There are costs associated with all that. There are also inefficiencies in grid computing, e.g., you may not get your CPU time at exactly the moment you wanted it.
2. $1/hour is $86,000/year, which is two orders of magnitude more than the cost of a headless cpu that you throw away a year later.
3. Sun's business prospects look uncertain, and nobody wants to commit themselves to a ship that may not be afloat in 10 years.
4. Many potential customers are military types, who would have security concerns.
5. Many potential customers are academic types, who may already be getting the job done with systems like seti@home.
6. Many potential customers are engineers, who think it sounds like a lot more fun to run their own machines.
7. IT managers would like to expand their feudal domains rather than outsourcing their work and therefore losing power, prestige, and staff. If the computation was being done off-site, the next obvious question for the managers' bosses would be why shouldn't they cut in-house computing resources.

Re:lots of reasons

[GGardner]

$1/hour is $86,000/year, which is two orders of magnitude more than the cost of a headless cpu that you throw away a year later.

You want to check that math again?

Re:lots of reasons

[nharmon]

You can get a low-power headless server system for $860.

Re: Lots of Reasons

[GogglesPisano]

Hardware is only half the battle.

I've been writing software for computing grid environments for about three years. It's an architecture that works well for problems with many discrete, parallel, CPU-intensive calculations (think Monte Carlo-type simulations). You can't just take any application, throw it on the grid, and expect it to scale up - the software really needs to be designed for a distributed environment. Issues of data movement, process synchronization and fault tolerance are all magnified when the application is spread across hundreds (or thousands) of machines simultaneously.

The companies that use these applications tend to be very large corporations (in my case, financial institutions, but I'm also aware of some oil companies that use them). Given the amount of development and IP required to build such systems, and the general tendency of such places towards paranoia, I believe they'd be reluctant to allow a third party to host these apps.

Re:Wouldn't it be easier...

[steve_l]

Depends on whether or not you value your data. Things like Seti@home and open blender can pull it off because their tasks are cpu-intensive, public data and nobody cares about timeliness of results. Also their algorithms are very cpu-intensive compared to the amout of data sent round.

If you have large quantities of private data, and your algorithm doesnt distribute as well as the massively-distributed examples, then you have less choice. You may want a large supercluster with infiniband backbone and a few terabytes of RAID-5 network storage.

It's plain too expensive

[Stephan Schulz]

I'm a potential user. I have a program that has an infinite number of strategies to tackle hard search problems, and a benchmark library of 8000 problems. Much of my work is to find out which strategies work well on which classes of problems.

I'm currently using our university student lab. But this is a mix of various machines, from 300MHz Sun Ultra 60 to 900 MHz SunFire machines, some of them limited in memory, and all used by students for their own nefarious purposes (e.g. pr0n and Quake). I'd love to be able to set 100 or so identical processors to the job. I could keep them fed for months. But at $1/CPU-hour, a day on 100 machines is $2400. I can buy 6 low-end Athlon machines for that money (and they will be just as good for the job). Yes, I do save in electricity and administration, but these costs are a) low for my application and b) come out of other budgets. For scientific work, SUN's prices are not acceptable. I would be tempted at a price of 1ct/CPU-hour. I would immediately buy into the thing for 0.1ct/CPU-hour with low-priority (i.e. I get to use only otherwise free processors).

Re:It's plain too expensive

[JesseMcDonald]

If the thrift stores go empty, people will buy new clothes.

Re:It's plain too expensive

[vinn01]

If people will buy new clothes, some will think:

1) Buy clothes that are no longer new
2) Sell them
3) Profit
4) Name business "Thrift Store"

... funny how business works.

Maybe Sun should study this model:

1) Sell cheap low priority CPU time on a grid system that nobody is using
2) No profit
3) Raise prices until the customers/costs/profits are balanced to favor profits
4) Profit

Re:It's plain too expensive

[Minna Kirai]

3) Raise prices until the customers/costs/profits are balanced to favor profits
4) Profit

That's not "Profit", just "Revenue". Profit would be if the income exceeded the investment they'd made in the Grid computers, which won't necessarily be the case.

All that approach guarantees is a partial recoup of sunk costs- and it might not even do that, if the salaries of the marketers and engineers who operate that service exceed the income.

However, drastically dropping prices would still be a good idea. If Sun wants any hope of their Grid service getting off the ground, they need to build up some customer volume- create a critical mass of IT people who are comfortable running jobs on a mysterious external Grid. Then later they can increase prices to the level of profitability, and hope the users are willing to stick around.

Might not work, but it's the only possibility.

Re:It's plain too expensive

[starfishsystems]

If you're an academic researcher or student, chances are that you can arrange access on one of the many clusters that are available to the research community.

If you're a Canadian researcher, for example, you can get a free account on WestGrid, simply in return for acknowledging its contribution in your published work, and for periodically reporting that research to WestGrid so that it in turn can build a case for continued funding. One of the Principal Investigators at WestGrid happens to be specifically interested in search.

In the United States there is TeraGrid and many other excellent cluster facilities, and so also in a number of other countries.

Researchers do not need to pay for commercial solutions which are only trying to catch up to what the research community has already been doing for several years!

IBM

[truthsearch]

I think it's more like they saw IBM making money from it and decided they can't let IBM have the whole market. It's not like this was Sun's idea in the first place. IBM was the first big company to make an offer. It seems to be working for IBM, so Sun already knew the market was there. What they probably lack is a clear differentiating factor for their customers.

Demands big customers

[Craig Ringer]

Well, it's very expensive for people who're not in a hurry.

If you need a job done *now* and don't have time to build the cluster, I can see how it'd be attractive. Especially if it scaled really well, so you could spread it out over 2000 machines and finish in a day, instead of 100 machines for, say, two weeks.

I personally wonder if Sun might've got more business if they scaled the prices, so putting jobs on a few machines was way cheaper than putting them on a lot. Maybe then they'd have got people trying it out. Right now, it looks like to be worth it you've got to be in a serious hurry and/or unprepared, and willing to bet on a service you haven't used before. hmm.

Another barrier might be comms. You still have to get the data to the "grid" and get the results back. If we're talking terabytes - which isn't unlikely when talking about buying tens of thousands of CPU hours - then that's a big whack of data. Combine that with a customer who's in a hurry and doesn't already have their own HPC infrastructure...

The reason as I see it

[kilodelta]

I think they really screwed the pooch on the grid concept.

Because all the folks I know who need the capacity need it for more than a few cycles. For example, friend of mine kicked Sun to the curb and built a 125+ cluster of Dell 2850's running Linux. This cluster does oceanographic simulation btw. So it wouldn't have been a candidate for Grid because it would have been hideously expensive, more so than the 125 computer and the necessary electrical and cooling improvements.

Ignoring small customers

[mnmn]

I was looking at their grid some time ago for some compilations. I wanted to build crosscompilers for other architectures (GCC ARM). There were several options and compilations took too much time including GCC, UCLIBC and the kernel itself, so I wanted to compile every iteration of the arguments in question (a selected list of options from the 3 packages).

Therefore I needed lots of CPU power. I browsed around their site, no way to just BUY something and start using it. I emailed them. They answered with something like $1000 per hour per CPU.... dont remember but it was just not worth it..

I got myself a remote server at serverpronto.com, at $30 per month with an Athlon CPU. Left compile batch files on it for a couple o weeks and I got what I wanted (wasnt a blazing success, but I got the files on the cheap). Next I bought an Athlon64 mobo + cpu and used my scsi disk on it to get the compiles much faster.

Sun has abandoned the smallscale hacker community and they still have a LOT to learn. Nobody will pay $1000 for a service theyve never used before, and dont know how to use. Setup server farms with Ultrasparc4 cpus and Linux/Solaris running and offer them for a few bucks an hour, or a $100 per month including high intenet bandwidth. Heck even I can make a profit from THAT.

Look up Sun in the dictionary ...

[LaughingCoder]

It says "irrelevant". Honestly, I posted a comment awhile back asking this forum if there was any more irrelevant company in tech these days than Sun. This article just convinces me even more.

Re:Look up Sun in the dictionary ...

[Miniluv]

Sun may be irrelevant in the larger picture (Sparc is a dead end, Sun kit is still absurdly expensive) but in the narrow OS world Sun is still highly relevant. Solaris 10 has really changed a lot of peoples minds on what they can expect from a proprietary UNIX, and should be reminding the Linux world why serious enterprises still don't take them entirely seriously. DTrace, ZFS, Zones, all of these are really excellent implementations of really cool ideas.

you are missing the end game

[handy_neil]

Don't over-think this. The end game for this campaign was the fill-page adds in the major financial newspapers that allowed them to re-assert as the premier scientific/engineering/serious business app platform.

The imagery of this grid of servers available on-demand is desined to stick to the brain of C-level decision makers.

I don't believe it!

[General Wesc]

You're telling me that Multics^H^H^H^H^H^H^Hthe Sun Grid Utility is a failure? But systems like this have always been commercial successes in the past! Just ask Bell Labs or GE.

oil industry, hollywood rendering

[peter303]

Seismic imaging and 3D computer graphics film rendering, are a couple of areas that might desire ten thousand nodes for a few months, and then dont want them for a while. I've heard of IBM offering a similar product. A major animation house rented a HP/Compaq super-cluster.

Not through lack of interest

[Anonymous Coward]

I've just finished the working-in-industry year of my degree, and the place I was working at was very enthusiastic about the Sun Grid; the problem was that they found it rather hard to port their (Linux and Solaris) software over to it. Don't ask me why, I wasn't in that team.

Re:[OT] Sorry for being your spelling Nazi

[donscarletti]

(really, I'm sorry, don't mod me to oblivion or flame me) it's Asperger's syndrome

Isn't being pedantic about language one of the sympoms of Asperger's syndome?

Aircraft designed and built by amatures[sic]

[phliar]

EAA.

Re:Aircraft designed and built by amatures[sic]

[RingDev]

Yup, and every year we get dive bombed in Madison. People joke about it, every year for the fly in you hear things like "If you live on the east side of town, its time to start wearing a helmet again". I can't remeber the last fatality free Oshkosh EAA fly in.

-Rick

Re:Aircraft designed and built by amatures[sic]

[Alioth]

There never will be a fatality free Oshkosh. TWENTY FIVE THOUSAND aircraft show up typically. It doesn't matter whether they are piloted by 100-hour private pilots or airline pilots - with that number of aircraft movements there are, you'll get the odd crash.

That doesn't mean the Oshkosh is a bad thing or that it should be banned. There are risks in life. It is exceedingly rare that an Oshkosh-related incident affects someone on the ground (such as the results of a mid-air collision landing on someone's house). It's not caused by the 'experimental, amateur built' status of *some* of the planes (a very large proportion, probably at least 50% of the aircraft are factory made, FAA certified aircraft - and the accidents that do happen are not limited to home builts).

Once a home built has flown off its initial 25 hour (or 50 hour depending on requirements) test period, most types have the same level of safety as FAA certificated aircraft.

No market research on the idea.

[crovira]

They never asked who the heck would want this. Then they could refine things like costs and opportunities.

This was an idea which was not required. The cheapest thing in the world is a CPU cycle. Unless you're doing things that demand far more that a Beowolf clauter can deliver, like SETI, and that aren't proprietary, like no commercial products I know of, you don't actually WANT this service.

What where they thinking?

Wow, thanks for that link

[Szplug]

I was considering building my own cluster, or possibly using Sun myself, but that company charges $1 per CPU per 24hr day; nice. And no power consumption worries either.

He he.

[hummassa]

You got me :-)

Easy

[Anonymous Coward]

85% / 90% of Office Resources is in Windows! How do you think I can format the HD from the raw BIOS with a VBA macro?

Here's a thought

[dtfinch]

Maybe $1 per cpu hour is too expensive. Either they'll have to sell the time at a market price, or shut the system down and write it off as a loss.

Re:A little more math, the actual customers...

[Wesley Felter]

Sun just wants about 5x what the market rate is for a CPU-year.

Not too surprising IMO. In order to keep latency low they may have to keep utilization low, which means they have to overcharge to make up for it. In general, renting something at small granularity is more expensive than renting at large granularity.

The article seems to be jumping the gun

[Anonymous Coward]

I have a friend who works at Sun, and just today he (and other employees) received an email from Schwartz saying:

In less than two weeks, Sun will officially unveil the world's first true computing utility. It's taken nearly a full year to build, and will be the only computing service of its kind.

So Sun hasn't officially opened up the grid for business yet. Given that, it seems par for the course that they wouldn't yet have customers.

Re:A little more math, the actual customers...

[Overzeetop]

The multiplier is too high, unfortunately. I have an issue like this, but it's for manpower. There are times during a year when I could use 10 architectural drafters per engineer, cranking away furiously for 2-3 days. Problem is, those times happen maybe once a quarter. I can't afford to hire even a single drafter full time and make money. If you could have "drafting on demand" from a pool of 1000 compentent drafters for double my normal cost ($25-$30/hr vs $12-$15) I'd jump at it. At 3x the cost, I'm scraping up against breakeven (remember - I can do it myself, in house, it just takes longer). At 5x the rate ($60-$75/hr), I can't bill out the service for what I pay for it. I'd rather take on less and have my lead times a bit longer than lose money on my biggest jobs.

Though you've been moded redundant, the core of your logic is right on topic - there's a limit to "need it right now" premium. Sun appears to have found a spot above that mark.

procestree.com

[Dwonis]

I'm not surprised. Anyone here remember processtree.com? It was a for-pay distributed computing company that never got off the ground, probably because it couldn't find any customers.

Logical decision process

[cameronpurdy]

First, you have to have an application that would benefit from Sun's "the grid". That rules out 99.9% of all applications right out of the gate. Then, you have to find one that isn't already running in some cobbled-together-yet-acceptable manner. That rules out 99% of the remaining 0.01%. You're left with two or three applications, tops.

Now think about the expense in actually preparing an application to run on this "grid":

- You have to re-build the app in a way that it can run on Sun's "the grid"
- You have to test it on "the grid" (this is probably the easiest part, and it still will cost at least US$100k in equipment, software and time)
- You have get the app bundled up in such a way that you can put it on Sun's "the grid"
- You have to get your data over there to Sun's "the grid" so that the app actually has something to crunch
- You have to get your lawyers to give you permission to put your data over there on Sun's "the grid"
- Your lawyers either just say "no" (probably!) or they suck up time from your best engineers to verify that it's possible to securely put your confidential data over there on Sun's "the grid"
- Your lawyers then say "no"

In the end, like most business decisions, we see companies going with the "known and trusted" instead of risking their assets on the "unknown and untrusted". Makes sense, doesn't it?

Peace.

Re:Logical decision process

[elmegil]

You seem to be of the misapprehension that I'm saying Sun's solution is better or best. I'm simply saying the grandparent poster is a moron who oversimplifies to the point of absurdity.

Privacy

[No Such Agency]

Of course it is interesting to see who (in the real world) are those companies?. If we suppose they are some top-notch companies that use a lot of processing power (like stock market companies wanting to run their models) they may preffer (and they may already have) to run their own servers to protect their secrets.

I suspect that one of the promises Sun makes is "your data and programs will never, EVER be at risk of theft, and will only ever be even known about by our machines, and one or two highly-trusted and superbly compensated employees". At least that's what I would promise...

Re:Not a question of price, but privacy, latency

[Decker-Mage]

This makes me think on the software liability issues point, I am sure these companies would demand something very very far from the typical "EULA" or contract to use this service. With services like this you demand a Service Level Agreement (SLA) which spells out exactly the level and type of services required, penalties for non-performance, and any other requirements (security of data, etc.) and their related penalties. In no way, shape or form do you have something that resembles an EULA where the provider can wave their hands and say "we aren't responsible." I have no idea what Sun is providing for their SLA. It'd be interesting to give it a read. Something else to go looking for if they publish it at all.

Re:[OT] Sorry for being your spelling Nazi

[Cervantes]

Oh, come on mods, kick up the two posts above this as +1 "Teh Funnay"!

Re:Flamebait indeed

[turgid]

...and sometime you're just a self-important asshole, and the mods recognize it as such.

Never underestimate the power of beer, young man.