Forgot your password?
typodupeerror
Government Supercomputing United States Science

What Kind of Cloud Computing Project Costs $32M? 158

Posted by timothy
from the those-investors-should-be-pissed dept.
coondoggie writes "The US Department of Energy said today it will spend $32 million on a project that will deploy a large cloud computing test bed with thousands of Intel Nehalem CPU cores and explore commercial offerings from Amazon, Microsoft and Google. Ultimately, the project, known as Magellan, will look at cloud computing as a cost-effective and energy-efficient way for scientists to accelerate discoveries in a variety of disciplines, including analysis of scientific data sets in biology, climate change and physics, the DOE stated. Magellan will explore whether cloud computing can help meet the overwhelming demand for scientific computing. Although computation is an increasingly important tool for scientific discovery, and DOE operates some of the world's most powerful supercomputers, not all research applications require such massive computing power. The number of scientists who would benefit from mid-range computing far exceeds the amount of available resources, the DEO stated."
This discussion has been archived. No new comments can be posted.

What Kind of Cloud Computing Project Costs $32M?

Comments Filter:
  • . . . but also the rest of the sky including the moon and the stars.
  • oh oh (Score:2, Funny)

    by Anonymous Coward

    "The number of scientists who would benefit from mid-range computing far exceeds the amount of available resources, the DEO stated."

    This sounds like one of those far-fetched statements that more realistically would be answered as "eleventy-billion."

  • The question is not "What kind of cloud computing project costs $32M?" The question is "Is research into the benefits of cloud computing worth $32M?"

    As with many multi-million research grants, it looks less like valuable research and more like a handout.
    • Re: (Score:2, Funny)

      by thefear (1011449)

      As with many multi-million research grants, it looks less like valuable research and more like a handout.

      Frankly I`m just suprised that the US government has a whole department dedicated to wasting energy.

    • I don't follow. Handouts are good. We are in a liquidity trap. We have massive unemployment and a 0% interest rate. Perhaps cloud computing isn't what we should spend money on. However, the $32 million those people get for building cloud computing will very likely be spent on what those people should spend money on. Until we can raise interest rates (due to improved employment), you are either pro-government spending on crap like this, or you are a gold bug. And if you are a gold bug, you should *still* be

      • I don't follow. Handouts are good. We are in a liquidity trap.

        I've never heard the term "liquidity trap" before, but yes, we do need the government to pump more money into the economy.

        However, there are much more effective ways, such as improving infrastructure (see 1930s and the building projects). More jobs are created for the same amount of cash (although lower salaries as they are blue color; still means less people filing unemployment). Less money is spent on products made overseas (you make your concrete locally, but those server parts are coming from Asia)

        • Clearly we should do those things first. However, those things being a better expenditure do not make this thing a bad expenditure.

          • by Gorobei (127755)

            Pretty much right. Although there doesn't really need to be an ordering in time: dump money in infrastructure, public beautification projects, scientific research, social services, education.

            It all is net positive while inflation remains low. Some people (mostly the rich) will always complain that "their" tax dollars are being wasted, but the real waste is that skilled workers deskill (e.g. accountants cooking their own meals, or plumbers darning their socks.) The economy is not a zero sum game.

    • Re:Wrong question. (Score:4, Insightful)

      by megamerican (1073936) on Wednesday October 14, 2009 @03:30PM (#29748723)

      The right question is who cares when the NSA is spending $2 billion [nybooks.com] just on the structure for a building (1 million square feet big) to house computers which will do who knows what for signals intelligence. Not to mention another facility in San Antonio being built which will be the size of the Alomodome.

      Let's not care about that but nitpick over something ~1% the size and far less destructive to our liberties.

    • by Matheus (586080)

      It's a question that from my interpretation of the article is valid but off-base.

      They are not spending 32M on *using* a cloud. They are spending 32M to *build* a cloud. That is why they specified "thousands of nehalem processors" ... they would have no knowledge of the underlying hardware if they were just talking about buying time on an existing cloud.

      I could easily spend 32M on building my own cloud. Actually that wouldn't even build a very big cloud... at consumer prices (which we all know they wouldn

  • Government Spending (Score:5, Informative)

    by Swanktastic (109747) on Wednesday October 14, 2009 @03:01PM (#29748305)

    You know, usually I'm against most government spending programs. They tend to be a huge waste.

    But this... It sounds interesting and could actually benefit basic research- something this country sorely needs to support. My (perhaps incorrect) observation is that some groups like the DOE and DARPA tend to allocate funds to valuable research projects rather than pissing money away on terrible administrative database implementations. I guess I should keep in mind that the majority of DOE funding is used to build and maintain our nuclear weapons fleet.

    • by LWATCDR (28044)

      Actually this is very reasonable. They are building their own cloud instead of maintaining many departmental clusters.
      The cost is to build their own cloud that can managed and probably secured.
      That is why it costs so much.

    • by cetialphav (246516) on Wednesday October 14, 2009 @04:06PM (#29749209)

      The DOE and DARPA (and others) are huge users of HPC (high performance computing) applications. The have a vested interest in having the state of the art advance in parallel computing and so they tend to provide lots of research grants to fund that. They also routinely let outsiders use some of their computing facilities for the same reason (not all of their labs do classified work). There are many computing facilities that need enormous computing power as shown on the Top 500 list. [top500.org] But they are seeing that there are times where researchers need computational power, but not at such a large scale and not for long periods of time. If medium powered computational facilities could be made available to researchers cheaply and quickly, they would be widely used.

    • Your right on being against government spending programs. In fact, everything outside of maintaining national defense, diplomacy, and a few other well defined functions of federal (not national) government, are done so poorly that continuing to dump money into them is the definition of insanity. In fact, our Federal government is so bad at doing things its not suppose to do... that it cannot even do them within the frame of directly collected tax revenue (see deficit spending). For example, if we cut Social

  • by PhilHibbs (4537) <snarks@gmail.com> on Wednesday October 14, 2009 @03:04PM (#29748379) Homepage Journal

    ...would that be mushroom cloud computing?

  • $32 million? (Score:5, Insightful)

    by condour75 (452029) on Wednesday October 14, 2009 @03:15PM (#29748503) Homepage

    With that much money they could get a quarter of an F-22 fighter jet! How dare they spend it on research?

  • by R2.0 (532027) on Wednesday October 14, 2009 @03:17PM (#29748545)

    The kind where the company who receives the contract is located in a particular Representative's district.

  • I imagine a large portion of that cost are salaries.

    • by HogGeek (456673)

      Obviously you don't work for the U.S. Government or one of its contractors...

    • by SETIGuy (33768)
      Assuming $5M is hardware and purchased software, the remaining $27 million would get you about 150 FTE years at government salaries, benefits, and overhead rates. Assuming the project is 3 years in duration, that's 50 FTEs, which seems high for doing something of this scope. Maybe they are planning to build a building to house it.
  • The trouble with supercomputing is that, if you have to share the thing, you don't need it.

    Supercomputers are worth the trouble if there are applications that need hours or days of time. But if you have many users sharing the thing, it's a waste. Price/performance tends to be maximized towards the upper end of mainstream machines. Supercomputers, with their custom hardware, tend to have lower price/performance than commodity machines. That's why web farms are made of commodity hardware.

    • There are problems which really need high memory bandwidth and don't fit on smaller-than-super computers, so a time slice on a supercomputer can be worth far more than full-time access to dedicated fast conventional computers. But those problems become less and less common as regular computers get bigger and faster - your laptop probably has a graphics processor that's faster than a Cray-2 by now...

    • The trouble with supercomputing is that, if you have to share the thing, you don't need it.

      You are misinformed.

      Most computer use, whether at a single workstation or a supercomputer cluster, is extremely bursty. A given user wants rapid turnaround for any particular computation, but otherwise leaves the system completely idle.

      This condition explains the success of consortia like TeraGrid and WestGrid which make vast computational resources available to a large user population. Those users could no
  • What's the bright side of cloud computing?

    When the cloud goes down, it's a bright and sunny day.

  • $32,000,000... (Score:3, Insightful)

    by corychristison (951993) on Wednesday October 14, 2009 @03:32PM (#29748753)

    ... sounds like a walk in the park compared to their other spending. I think that number is off by a factor of 100 or so.

    In contrast, my small city (~40,000 people) in central Canada is spending ~$56,000,000 on a new Multiplex/Sports center. Supposed to have a new hockey rink, curling rinks, soccer area's with artificial turf.

    I'd my city council spend it on a Cloud Computing Centre.

    • by CannonballHead (842625) on Wednesday October 14, 2009 @03:35PM (#29748803)

      I'd my city council spend it on a Cloud Computing Centre.

      And schools to [intentionally left blank] how to use verbs in sentences.

    • by lymond01 (314120)

      Not that basic research isn't a great thing, but a sports center, properly funded and located, generates income and doesn't need to be recycled every 3-4 years.

      But I bet you'll find cheaper hot dogs at the Cloud Computing Center. Or centre.

    • by geekoid (135745)

      How much will the small city get from the center?

      That's the question. It's not how much money they spend it's what value does it have compared to the money.

      • How much will the small city get from the center?

        That's the question. It's not how much money they spend it's what value does it have compared to the money.

        Virtually nothing.

        There is a much larger city 65km (40 miles) away that has much more to offer in terms of facilities and services.

        Our new facility will not be able to accommodate any type of concerts, so they'll go to the other city... that also has the airport (next closest is 225km away).

        Just seems like a waste of money. We do need a new Hockey and Curling facility (but have many outdoor soccer facilities). I think our council could trim some bulk off that $56 million and put it towards other things.

        Oh,

  • Beowulf cluster of....

    Oh, wait...

    Never mind.

  • The specifications for that cloud include a silver lining.
  • is smoke, and the project was titled "Burning 32 millons"
  • by DutchUncle (826473) on Wednesday October 14, 2009 @03:42PM (#29748905)
    When the last ATC project failed disastrously, people were already playing online games over phone modems. Now we have massively multiplayer games, with gigahertz hardware dedicated to each user (your PC, that is), and ATC is still being done on single mainframes. Quick scan suggests six thousand planes in the air at a time over the US; let's call it ten thousand. Dedicate a CPU to each plus some hierarchy of busy areas and regional control; allow $1000 per CPU/system (and its share of comm bandwidth); call it $10 million. Sounds like an interesting project. :-)
  • We already have a platform to do this - BOINC. We've been wasting megawatts on SETI for years. Perhaps we should turn the search closer to home and just search for terrestrial intelligence, but that could be equally futile.
  • Easy: the one where you are building the cloud.

    Makes sense to me!

  • As the saying goes: keep your feet on the ground instead of your head in the clouds.

  • ..although for $32 million, this cloud probably has a Tantalum lining!

  • by awitod (453754)

    What kind? The kind that requires a building that sits on land and is full of hardware.
    $32 million isn't that much when you consider that.

    Here is an estimate for an empty 80000 square foot office building with no contents and no land. ~$12 million.

    http://www.reedconstructiondata.com/rsmeans/models/offices3/ [reedconstructiondata.com]

  • 32M sounds like a giant chunk of change, but, its not even what gets spent on FireFox each year.

    You figure a 50 man team of senior devs for a year, and I think that would pretty much do it.

  • 'what kind of cloud computing project costs $32 mil' it says.

    hell. even the bare costs of the number of servers that would be required to run a cloud of that size would amount to a goodly portion of that $32 mil. EVEN if you buy them in bulk. thats leaving out everything else including the datacenter setup, software, administrators, engineers, the team to create the project.

    doh. i suppose cloud computing comes free, in the universe the article poster lives in.

How many QA engineers does it take to screw in a lightbulb? 3: 1 to screw it in and 2 to say "I told you so" when it doesn't work.

Working...