Slashdot Mirror
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."
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.
If you "get" pointers add me as a friend (116)!
Obviously this news article is about as popular as SUN's new program
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?
Oh, say does that Star-Spangled Banner entwine / The myrtle of Venus with Bacchus's vine?
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....
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?
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.
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.
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).
If you "get" pointers add me as a friend (116)!
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.
Join the Empire! http://www.empirereborn.net/
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.
Test your net with Netalyzr
"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
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.
You're nothing; like me.
... 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.
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.
My blog
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.
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...
Because, after all, those machines are all self-maintaining and configuring.
7 November 2006: The day Americans realized corruption and incompetence weren't addressing 11 September 2001
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.
Find free books.
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).
Stephan
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.