Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Power Cloud Earth

Energy Consumption at Data Centers Will Become 'Unsustainable', Researcher Predicts (www.cbc.ca) 121

"The gigabytes of data we're using -- although invisible -- come at a significant cost to the environment," argues the CBC's senior technology reporter. "Some experts say it rivals that of the airline industry." And as more smart devices rely on data to operate (think internet-connected refrigerators or self-driving cars), their electricity demands are set to skyrocket. "We are using an immense amount of energy to drive this data revolution," said Jane Kearns, an environment and technology expert at MaRS Discovery District, an innovation hub in Toronto. "It has real implications for our climate." [Kearns is also the co-founder of the CanadaCleantech Alliance....]

It's not the gadgets themselves that are drawing so much power, it's the far-flung servers that act as their electronic brains... The data centres, often bigger than a football field, house endless stacks of servers handling many terabytes (thousands of gigabytes) of digital traffic. Just as laptops tend to warm during heavy usage, servers must be cooled to avoid overheating. And cooling so many machines requires plenty of power. Anders Andrae, a researcher at Huawei Technologies Sweden whose estimates are often cited, told CBC News in an email he expects the world's data centres alone will devour up to 651 terawatt-hours of electricity in the next year. That's nearly as much electricity as Canada's entire energy sector produces. And it's just the beginning.

Andrae's calculations, published in the International Journal of Green Technology, suggest data centres could more than double their power demands over the next decade. He projects computing will gobble up 11 per cent of global energy by 2030 and cloud-based services will represent a sizeable proportion of that. "This will become completely unsustainable by 2040," Andrae wrote...

The information and communications technology sector as a whole is thought to be responsible for two to three per cent of global greenhouse gas emissions -- roughly on par with the often-criticized airline sector.

The article also notes that by 2018 Amazon had already switched to renewable energy for at least 50% of its cloud computing servers. And it also adds that tech companies "in Canada and abroad...are coming up with innovative solutions to curb the growing problem."

Solutions being investigated include recovering the low-level heat generated by servers and increasing the speed and capacity of data transfers (including one approach involving a quantum dot multi-wavelength laser).
This discussion has been archived. No new comments can be posted.

Energy Consumption at Data Centers Will Become 'Unsustainable', Researcher Predicts

Comments Filter:
  • i’d be more worried about skynet.

  • by bluegutang ( 2814641 ) on Sunday January 05, 2020 @03:56PM (#59589702)

    Using lots of electricity is not actually a problem. There are many ways to generate carbon-free electricity, the most reliable being nuclear fission. Build a new nuclear power plant and it can supply power to any number of data centers you want to throw at it. Or else, put the data centers in Iceland where geothermal energy is abundant and the outdoor weather helps with cooling too.

    Airplanes are a climate change issue because they have to carry their fuel with them, and pretty much nothing is as energy-dense as burnable hydrocarbons. But for a data center which is just sitting there in a location of your choice, there are many forms of carbon-free electricity to choose from.

    • Re: (Score:2, Insightful)

      by gweihir ( 88907 )

      Nuclear fission heats up the environment rather directly by its massive cooling needs. Also, it is massively more expensive than any other power sources. Pouring tons of money into it when clean energy is available a lot cheaper is the height of foolishness.

      • eh? the whole point of a nuclear reactor is that heat gets turned into electricity by boiling water through a turbine. There's very little excess heat that gets out as even the excess is used by efficient systems.

        the cooling you might be thinking of, is actually the water required to grab that heat to turn into electricity.

        • by gweihir ( 88907 )

          You seem to be unaware of basic physics. The Carnot cycle needs all that steam to be cooled down after the turbine to the previous temperature of the water to work.

          Typical heat engines get 30-50% efficiency, the rest is dumped and heats up the environment.

      • How much power do those solar panels produce at night? How much power is there from a windmill without any wind?

        There is no future without nuclear power. Japan discovered this, and Germany is learning this lesson too. France is going to discover this, and Australia is finally seeing how nuclear power is a good idea.

        The height of foolishness is seeing wind and solar power continue to fail to replace coal and natural gas and thinking if only more money was dumped in this money pit that will solve the problem.

        Batteries won't solve the problem. Australia already saw that while batteries are useful for managing the production of electricity from wind power it can also manage the unscheduled shutdown of a coal fired power plant. In the future I expect to see large batteries installed next to nuclear power plants to manage changing electrical demand, the nuclear power plant will just hum along keeping that battery charged up. If there's a shutdown, planned or not, then the battery will fill in until corrective measures can be taken. Just like what happened in Australia with that large coal plant.

        The future is onshore wind, hydro, geothermal, and nuclear fission. Oh, and batteries. The past has demonstrated the inability of a wind and solar only policy to be in our future.

      • Nuclear fission heats up the environment rather directly by its massive cooling needs.

        Nope, this is not really relevant to climate change. When you burn fossil fuels, every ton of CO2 released will stay in the atmosphere and heat it for decades or centuries, long after you shut down the plant. A nuclear plant only produces heat while it's running.

  • game streaming is bad as that needs lot of places right near each user base.

    • My initial reaction is that I don't know why that would be, where is the increment in energy cost vs the same stuff running at home. You could argue the recycling of the hardware to service different users at different times vs perhaps being left on idle at home maybe a saving. (No idea if any of the services properly shutdown idle instance only having a small pool ready but idle at a time?) Further thought I guess it depends on time of year and location. In cooler places at the right time of year I guess
  • Isn't that fact that we're obviously going to only increase the amount of energy we use proof enough that we should be investing in nuclear Small Modular Reactors? If you need lots of power all the time then there is no better source than nuclear energy. Do people not think that as we move from ICE cars to battery EVs that we are going to use a shitload of power at night charging those EVs?

    "Nuclear is super expensive!" some dumbass is going to claim while completely ignoring that SMRs are specifically des

    • Re: (Score:2, Redundant)

      by careysub ( 976506 )

      "Nuclear is super expensive!" some dumbass is going to claim while completely ignoring that SMRs are specifically designed to be inexpensive...

      Since there are no operating SMRs anywhere, it is a little too soon to declare that they are inexpensive as claimed.

      Remember the Gen II+ reactor designs, like Westinghouse's AP1000 [westinghousenuclear.com]?

      They were also designed to reduce costs and were promoted as being "economically competitive". There are four such units operating in China, built by China, but of nine units started in the U.S. only two will be completed (Vogtle Units 3 and 4), the others being cancelled because of cost, and Vogtle units went far over budget -

      • by blindseer ( 891256 ) <blindseer@noSPAm.earthlink.net> on Sunday January 05, 2020 @06:30PM (#59590180)

        Remember the Gen II+ reactor designs, like Westinghouse's AP1000?

        Are you arguing that because a 40 year old design doesn't meet your expectations that we should not continue with new development? Tell me something, what was the state of solar power 40 years ago compared to today? I can imagine a lot of development and improvements were made with solar power. I seem to recall a lot of new houses around here getting solar panels on their roof lately, including my brother's house.

        Could it be that solar power made such gains because the government allowed for people to construct new solar power projects? In fact I recall the US federal government, and many state governments, giving large subsidies for solar power. That might explain the rapid pace of development.

        Just think how nuclear power would be today if the government took even the slightest interest in allowing nuclear power plants to get built for the last 40 years. We can do a lot of simulations. There's non-nuclear scaled models that can tell us a lot about thermal and chemical properties. But nothing tells us how a new design will operate like a full scale and operational prototype.

        Since there are no operating SMRs anywhere, it is a little too soon to declare that they are inexpensive as claimed.

        And we will never find out unless we start building them. You want the nuclear power advocates to learn just how expensive it will be? You want to demonstrate that nuclear power is simply too expensive to bother? This will forever be an unknown until someone actually tries. Because nobody was allowed to build anything new in the USA for 40 years we lost a lot of time in getting better nuclear power, or finding out that it won't work.

        Due to lack of transparency in China, and peculiarities of Chinese conditions (lack of environmental controls, etc.) it is hard to evaluate their actual cost, though maybe we could China to build us some and see.

        That's what I expect will happen if the USA doesn't do something about their backward rules on building new nuclear. We will just have to wait until China, Russia, or India opens some new nuclear power plants, proving their viability. We are already seeing new nuclear power plants being built in Japan and South Korea but for some reason those don't carry the same kind of weight as when long time adversaries do this.

      • At the current rate of consumption, the amount of uranium on the planet is only good for another 200 years [scientificamerican.com]. Using enriched uranium and other measures will extend that time. If we could extract the uranium from sea water we'd have 60,000 years of electricity. Go with breeder reactors and you're set for 30,000 years at current consumption rates using available uranium.

    • SMRs are specifically designed to be inexpensive

      ^ This was the most facepalmy part [wiseinternational.org] of that whole drivel of yours:

      A 2015 report by the IEA and the OECD NEA predicts that electricity costs from SMRs will typically be 50-100% higher than for current large reactors

      • by sfcat ( 872532 ) on Sunday January 05, 2020 @05:45PM (#59590024)

        A 2015 report by the IEA and the OECD NEA predicts that electricity costs from SMRs will typically be 50-100% higher than for current large reactors

        They had their lawyers contact the environmentalists ahead of time to fixed rate price for the lawsuits then? Let's see here, a 250 MW SMR costs ~ $15m to produce and about the same each year to operate (at most). And it produces $160,000,000 / yr in revenue at $0.02/kwh with 1/25th the waste of current reactors and that waste only remains radioactive for 300 years.

        Yeah, that's totally not going to be cost competitive. The cost of nuclear is the cost of the lawsuits. Until environmental lobbying groups support nuclear, they have no creditability. Hope you enjoy your fracking caused flammable tap water. But yeah, tell us again how we will power everything with solar and wind again despite all evidence to the contrary. Tell us how we will be able to ramp up our battery production by 1000x without any environmental costs. Tell us about how a reactor that can not meltdown is dangerous despite all evidence and data on the topic.

        • by blindseer ( 891256 ) <blindseer@noSPAm.earthlink.net> on Sunday January 05, 2020 @06:48PM (#59590232)

          But yeah, tell us again how we will power everything with solar and wind again despite all evidence to the contrary.

          Evidence like this?
          https://www.washingtonexaminer... [washingtonexaminer.com]

          More than 1,000 additional people are dying each year from air pollution due to Germany's phase out of nuclear power in response to the 2011 Fukushima disaster, according to a new study.

          Those deaths are attributable to coal-fired power predominantly replacing shuttered nuclear power plants in Germany, driving around a 12% increase in local air pollution. That rise in pollution alone bears a price tag of $8.7 billion per year, more than 70% of the phaseoutâ(TM)s total annual costs of $12.2 billion, economists at the University of California at Santa Barbara, University of California at Berkeley, and Carnegie Mellon University determined in a paper circulated Monday by the National Bureau for Economic Research.

          The paper, which has not undergone peer review, follows another recent report finding more people died from the shuttering of nuclear power in Japan after the Fukushima disaster than from the accident itself.

          Japan and Germany tried to phase out nuclear power, and they are seeing more deaths than if they just kept the nuclear power plants operating.

          This is just a precious gem too...

          Nuclear power, for example, is the largest source of carbon-free power in the United States. Some Democratic presidential candidates, however, such as Vermont Sen. Bernie Sanders and Massachusetts Sen. Elizabeth Warren, have raised significant questions about the role of nuclear power, despite having robust climate change plans. Warren recently walked back her comments somewhat, suggesting sheâ(TM)d keep some existing nuclear plants online in the interim.

          Now we see politicians like Senator Warren walking back from demands to shutter all the nuclear power plants. Is nuclear power dangerous and expensive? If so then why would we not close them all down as soon as possible? We know why, because nuclear power is very safe and affordable. By keeping them open we aren't being forced to turn to coal to keep the lights on like Japan and Germany had to do.

          Here's the problem though, those nuclear power plants will wear out. It might be another 20 or 30 years but they will need replacement. If we don't start building new nuclear power plants now then we will simply fall into the same trap Japan and Germany already did.

          These long time Democrat politicians just painted themselves into a corner on nuclear power. They know we can't supply our future energy needs with wind and solar power alone. We will need hydro and nuclear too. By speaking out against hydro and nuclear for so long they will have to tread very carefully or they will end up like the fools they are. What we are seeing now is a new generation of Democrats, with one group keeping this nonsense and another realizing the failed policies of the past. Democrat primary voters will have to choose wisely, because keeping the same old energy policies will mean more deaths and higher energy prices.

          The future is onshore wind, geothermal, hydro, and nuclear fission. We can use the evidence before us now to realize this or we can fail to learn from history and repeat the same mistakes.

          • by Bongo ( 13261 )

            Wind is like a gateway drug for new nuclear. It normalises the notion that massive energy generation can be "clean". Which is just an old moral thread about "purity". Anyway, pretty soon the novelty of wind turbines will wear off and people will become sick of seeing them. Just like we got sick of seeing too much car traffic and too many planes in the sky. And at that point we'll want clean and out of sight. Some clever industrial designers will get involved, to make sure they don't look like traditional 3

        • What is it about lawsuits? Everyone knows that small reactor units are expensive per kWh and bigger will always be better. That's the whole reason why they got bigger over time, which you apparently didn't even notice.

          Let's see here, a 250 MW SMR costs ~ $15m to produce

          Like WTF? In what fictional universe do you live in? The latest small nuclear power plant is the Russian 2xKLT-40 unit with 70 MW of output and an estimated price tag of around $450M [themoscowtimes.com] - which is about 100x more expensive per MW than what you claim. Again, there's a fucking reason why if you go

    • by mspohr ( 589790 )

      The problems with nuclear is that it is expensive and requires long lead times... plus still no good solutions to nuclear waste.

      However, wind and solar are cheaper than even coal and natural gas (and much much cheaper than nuclear) and can be deployed quickly so no reason not to use as much renewable energy as we want.

      • Re: (Score:2, Informative)

        by blindseer ( 891256 )

        The problems with nuclear is that it is expensive and requires long lead times... plus still no good solutions to nuclear waste.

        However, wind and solar are cheaper than even coal and natural gas (and much much cheaper than nuclear) and can be deployed quickly so no reason not to use as much renewable energy as we want.

        If this was true then why all the concern about "unsustainable" energy problems in the future?

        There's concern because this is not true. We know how to dispose of nuclear waste safely and at low costs. Nuclear power is cheaper than solar and offshore wind. The lead times on building new nuclear are from bullshit lawsuits and inexperienced workers. We were able to go from groundbreaking to producing power in less than three years with nuclear power decades ago. What happened? We lost a lot of experience

      • by gweihir ( 88907 )

        Indeed. And that is the rational stance here. Nuclear is a failure at this time, and making it viable (if that ever happens) will take far too long.

    • by gweihir ( 88907 )

      "Nuclear is super expensive!" some dumbass is going to claim while completely ignoring that SMRs are specifically designed to be inexpensive and completely ignoring the external costs of other power generation methods (especially fossil fuels).

      And then, there are no working prototypes and the nuclear mafia has a history of massively over-promising and massively under-delivering, especially on anything cost-related.

      "Nuclear waste is forever!" likely the same dumbass is going to claim while ignoring the fact that there are types of nuclear power that produce no waste and uranium based SMRs produce far less waste and the thimbles full of waste that they do produce is safe after 200 years.

      And there I was actually researching the issue and finding this is a direct lie. Sure, Uranium SMRs are more fuel effective, but the waste they produce is still a problem for a few million years.

      The dumbass here is you.

  • It's the bitcoin (Score:1, Insightful)

    by Anonymous Coward

    Many data centers are used and aboused to mine bitcoin with all the unused hardware. Guess what suckes power and returns nothing to humanity except a dminishing return of profits to speculators? And guess why AWS just started supporting "arm64"? Because employees and hackers keep stealing credentials to build up huge networks to run botcoing. It's proably 80% of most AWS servers, especially including ordinary "servers" that have been rootkitted to *also* run bitcoin.

  • Who the frak needs an appliance that tells you what you need to buy of if you have only a few eggs or vegetables left?
  • The information and communications technology sector as a whole is thought to be responsible for two to three per cent of global greenhouse gas emissions -- roughly on par with the often-criticized airline sector.

    In other words, it contributes bugger all. All that statement tells me is that the actual pollution by airlines is in no way proportional the the volume at which certain people are screaming about it. If we stop flying and close all of our data centers, we will not have solved global warming according to the prevalent models.

    • That's like saying, "You've got skin cancer and lung cancer. We're going to ignore the skin cancer because lung cancer is the big killer."

      This is fallacious thinking at its best. There is no reason why we can't make data centers more efficient at the same time as we make transportation more efficient.

      No, picking the small one isn't sufficient, but there's nothing saying that we can't tackle more than one thing at once. And tackling the small stuff isn't without merit. Every little bit helps.

      • Not quite the same. It's not the CO2 from flying or data centers specifically that's going to do us in. We need to reduce CO2 output by 50% (or whatever the figure is) but the reduction can come from anywhere. But each of those types of cancers is dangerous on its own, treating one will not eliminate or even reduce the risk of death from the other.

        Yes, every little bit helps. If we can slash emissions from data centers in half with some easy measures like adding solar panels and improving thermal eff
      • That's like saying, "You've got skin cancer and lung cancer. We're going to ignore the skin cancer because lung cancer is the big killer."

        No, it's like a physician telling a patient he's got a vitamin D deficiency. There's several ways to correct this. It can mean getting more sunlight. It can mean eating more fish or more mushrooms. Perhaps getting it from taking a pill, or drinking fortified milk. For patients some of these options will be more appealing, more effective, less in cost, or less inconvenient. They will have to decide which is the best means to reach the desired result.

        What's the problem again? It's CO2 production. Wher

  • I pledge to help the environment by unplugging my "smart" refrigerator.
  • by VeryFluffyBunny ( 5037285 ) on Sunday January 05, 2020 @04:45PM (#59589826)

    We're unlikely to make much of an impact on global heating by addressing power consumption in data centres. Yes, they produce wonderful feel-good stories about using solar & wind & building them in colder places to make cooling more efficient. But we're talking about effects on less than 3% of total CO2 emissions which constitute less than 1.5% of total greenhouse gas emissions.

    The biggest impacts will come with reducing emissions from buildings & cars.

  • by Lije Baley ( 88936 ) on Sunday January 05, 2020 @04:55PM (#59589878)

    The elephant in the room, cough*advertising*cough, drives so much of the wasteful behavior in web apps, on both the server and client end, all burning way more cycles than needed to catch attention, monitor, engage, and stretch everything. All the extra scripts and files and extra code in each of them. I don't know what practically can be done about it, but it would be nice if the "climate conscious" folks would go after that instead of trying to get my gas furnace. The internet and modern apps are rife with bloat and waste and reinvention, all which waste energy, both directly and through all of the churn involved in (re)making them.
    OK, rant over, I'm going back down to the senior center to play bingo now...

    • by Calydor ( 739835 ) on Sunday January 05, 2020 @07:06PM (#59590304)

      It's not just ads. Sitting on a really slow connection it is becoming obvious that sites absolutely do not care about optimizing anything anymore, they'll just dump tons of data whether it's required or not.

      Case in point, I was looking up some lists of gear for an MMO today. Just a list of the optimal gear available. Even with descriptive text that should be no more than 10 KB.

      It took FIVE MINUTES for the page to load, which means it's roughly 10-15 MB. For a list of items. Not even with pictures.

      • by Lije Baley ( 88936 ) on Sunday January 05, 2020 @08:00PM (#59590412)

        Indeed, the more you think about it, the uglier it gets. I saw the comment on game streaming and was thinking about video streaming. I would love to see a comparison of how much total energy, on all ends, is being used to enable the luxury of mass individual video streaming compared to broadcast television.

  • I saw them for the first time in the late 90s. They were ridiculous then and, for the most part, they remain ridiculous now.
  • Andrew McAfee on More from Less [econtalk.org] — 14 October 2019

    Andrew McAfee: One of the things that I found, that I put in the book, was this wonderful article by a retired newsman in Buffalo. And he loves to go around garage sales and just look for historical artifacts in Buffalo. He bought a stack of used, of old Buffalo newspapers from the 1990s and he found a Radio Shack ad from, I believe, 1991.

    And he said: 'You know what's interesting about this ad? There are fifteen gizmo-type items on it,' he said. 'Thir

  • Not new news (Score:4, Informative)

    by smoot123 ( 1027084 ) on Sunday January 05, 2020 @05:49PM (#59590034)

    I seem to recall hearing discussions in, oh, 2007? about how power density in data centers was becoming a huge issue. That's about when CPU development went from "crank the clock rate" to "crank instructions per watt". It's also when data center design and efficiency became an issue. People started talking about the ratio of power used for compute vs. cooling (and in typical cases, that was less than one).

    So I'm not certain why this is news today. It's an issue we've known about for a long time and lots of smart people are beavering away at mitigating it.

  • by SuperKendall ( 25149 ) on Sunday January 05, 2020 @05:49PM (#59590036)

    Luckily for the world, for some time now I've been working on the idea of a Bit Windmill, that should generate a much power as data centers need.

    The way it works in ingenious - you know how on an oscilloscope, a 1 is a high waveform, and a 0 is a low? Well, the bit windmill is placed in between these states, so that as a 1 changes to a 0 or vice versa, it moves past and thus imparts spin on this very tiny "windmill" - the more bits change, the more energy you generate!

    The only change needed globally to insure maximum power generation, is for every web page to include the BLINK tag in the HTML so that the cycling background pixels will ensure a constant supply of energy from the bit windmills.

    • Why bother? My perpetual motion machine generates all the energy we'll ever need.

    • You could probably put a light-collecting panel in front of the monitor to collect the light after it is used, too. No need to light up the rest of the room, turn it into electricity!
  • Comment removed based on user account deletion
  • If we code better, we don't need to use the full capacity of every CPU and DIMM in the DCs. As such, we can use the same computing power and just be able to increase on demand as needed. What the article means is if everyone codes poorly it is unsustainable! if everyone codes well, the number of endpoints in the DC will decrease as computer power goes up, because each node will be able to scale up on demand as needed.

  • by mattb47 ( 85083 ) on Sunday January 05, 2020 @06:45PM (#59590230)

    Companies already place lots of datacenters in places with cheap electricity via hydro (e.g. Washington and Oregon). Or cheap geothermal (Iceland).

    Get even more hydro going in northern climes (Canada, Nordics, Scotland, Northern US) and South America (lots of rivers thanks to the Andes). And place more datacenters in those locations. Plus cold locations can pipe in outside air for cooling.

    This can be done without needing tons of coal/natural gas/etc.

    Note wind and solar are too sporadic to really work here. They don't provide baseline power.

    • The future is onshore wind, hydro, geothermal, and nuclear fission. These are energy sources that are already low in cost, low in CO2 emissions, high in energy return on energy investment, and low in needs for land and raw materials.

      You say that wind is too sporadic. I can agree with that but onshore wind is also too cheap to ignore. Wind when combined with inefficient natural gas turbines as backup is expensive and can often be higher in CO2 emissions than if the natural gas was just burned in a far mor

  • Article should have mentioned how much energy is wasted mining cryptocurrencies. That'd probably be #2 after Netflix/streaming.

    • Article should have mentioned how much energy is wasted mining cryptocurrencies. That'd probably be #2 after Netflix/streaming.

      Then there is the wasted energy in people posting comments on the internet. Like this one I am posting now. Think of all the energy wasted in transferring useless commentary like mine.

      And now I will click "submit"...

  • I'm pretty sure my streamed music, movie and interactive gaming is just about the least energy intensive activity per $1 of GDP. We are all getting richer. World GDP is growing at about 5%. We need more data centers, we need a bigger virtual economy.
  • disposable VM tech like docker, snap etc. are as bad as plastic bags and straws; they're very cpu/memory/disk wasteful compared to proper bare metal app deployment.

    Docker is cancer.

    • If you're trying to get Trump supporters to use Docker, it's working.
    • Virtual machines are just another kind of hardware abstraction. We have the computers we have today because of abstraction lowering the time and effort needed to develop software.

      I'm barely old enough to remember the days of MS-DOS. This was when people would exit the Windows UI to run their games "on the bare metal". This meant every game had to go through it's own hardware setup for video, sound, and keyboard/mouse/joystick/whatever. A new video card meant doing this all over again with every DOS game

    • Probably what makes them wasteful is some kind of market failure in the way that prices are calculated for cloud computing, where keeping a larger capacity VM running for 30 days is somehow more expensive than dynamically creating and destroying smaller VMs. Although I would be kind of surprised if the likes of Amazon haven't done that math and figured out how to embed the cost of dynamic VM creation/destruction to avoid customer VM churn or at least make it so slight its not worth the effort to do very mu

  • Could the data center buildings have a centralized liquid-cooling system for all the processor banks that is more energy efficient?

    Apologies in advance if this is a totally ridiculous idea/question. I don't have any experience in this apart from building PCs and the one time in my job I was allowed to walk around inside a data center building. All the processors in the data center were air-cooled, but it seems like sheer number and power needed (i.e. economy of scale) for cooling may be large enough to desi

  • Once upon a time people would hop in the car and go for a Sunday drive. Nowadays everyone has their nose glued to a screen instead. Just like you and me right now. Not spending the day driving around for an unnecessary purpose must offset some of that data centre power, especially when all cars are finally electric.
  • by Solandri ( 704621 ) on Sunday January 05, 2020 @09:05PM (#59590578)

    he expects the world's data centres alone will devour up to 651 terawatt-hours of electricity in the next year.

    This is the same mentality that thinks EVs are somehow zero emissions, as if the electricity to power them is magically created out of thin air. EVs simply shift the emissions to the power plant (which granted, if it's nuclear or renewable is in fact zero emissions).

    In the same way, it's disingenuous to add up all the electricity consumption by data centers as if they're a separate entity to try to make the number as big and scary as you can. How many people use or derive some sort of benefit form those data centers? Now attribute that electricity use to all those people - divide the electricity consumption over those people.

    • 651 terawatt-hours per year
    • OECD nations have a population of 1.3 billion [wikipedia.org]
    • 651 TWh / 1.3 billion = 501 kWh per person
    • Global average electricity price is 15 cents/kWh [globalpetrolprices.com]

    (501 kWh/person) * ($0.15/kWh) = about $75 worth of electricity per person each year.

    You don't even need to switch to powering the datacenters with nuclear or renewables like others have posted. Just put them in cold climates near where people would be burning electricity or fuels for heating anyway. Thermodynamically, that's all computers really are - a fancy electric heating element.

  • The trick is to use the waste heat for operations that need it. Like heating dwellings, heating greenhouses for hydroponics, etc.
  • The brain is also a big energy consumer in the human body. It's just natural.
  • NU
    CLE
    AR
    POW
    ER
    !

    • "Stirling Engine" Oh come on, what does a nuclear power plant run on? Not excess heat. A Stirling engine does. And stationary applications are best for them, as they are a bit heavy. You couldn't run the entire data centre on it, but it would be stupid to just throw all the free energy away.
      • by Chas ( 5144 )

        Please see "conservation of mass & energy".

        A Stirling engine might OFFSET some of the power required.
        But as a primary generation source? No. Sorry. The power density and the initial energy needed to get a Stirling engine going just aren't there.

  • Fred Pohl wrote a story once where global warming was basically just caused by the amount of energy being used by "clean" fusion plants - it all ends up as heat one way or another and TANSTAAFL.

  • by grumpy-cowboy ( 4342983 ) on Monday January 06, 2020 @09:11AM (#59591858)

    Until the trend is to trough more hardware to compensate for sloppy programming/architecture/deployment/management/... practices (you know the "RAM/HDD" is cheap mantra, Kubernetes, "web-scale" even if not required, ...) at lot of energy will be wasted. Reusing the heat is not the solution. It's just a band-aid on the problem. Cut energy consumption/waste at the root is the solution.

  • So long as BC and Quebec have that sweet sweet hydroelectric power, Alberta can't find a reason why we should be mining tar sands to make dilbit oil (sludge, really).

    So they're trying to prime the pump with fake news stories about energy crises that are never going to happen.

"The vast majority of successful major crimes against property are perpetrated by individuals abusing positions of trust." -- Lawrence Dalzell

Working...