The real benefit to a 100F setpoint is the free cooling it allows. You can use filtered out door air or evaporatively cooled water from a cooling tower to keep a datacenter at 100F year round just about anywhere. This is a 90% reduction in cooling energy right there using decades old, mature HVAC tech.
The main benefit is that it allows free cooling, either bringing in filtered out door air directly or using a cooling tower to generate evaporatively cooled water for direct use in coils. At a space temperature of 100F, you could cut out 90% of the cooling - even in a full-time humid climate like Malaysia. In most US climes, you can do better.
I'm sorry, but you still haven't shown any proof that this study needed to be done by the government. You can try to change the subject all you want, but that doesn't change the fact that any competitive data center would be aiming for higher efficiency because it directly lowers their costs and allows them to offer lower prices than their competitors.
This just is not true. I work in the industry, performing studies of this nature. The Uptime Institute, 24/7 Group, etc. are not enough to promote sharing of this data between rivals. In a similar arena (semiconductor cleanroom critical environments), I've been paid to do the same damn study three times - but the third and last time it was funded by the government (LBNL) and disseminated publicly (something industry groups, such as SEMATECH, do not do). That is efficiency that saves money on an industry wide scale (and I have plenty of work, I don't need to do the same thing over and over for every little group of companies out there). The government's investment saved industry ten, possibly a hundred fold, expenditure. That's good business on a national economy scale. Who else would be willing to invest in the economic commons?
Besides that, if the government isn't going to pass legislation based on the study results, why even bother?
The Energy Star program has proven across numerous fields the value of providing information to the market. Are you at all familiar with it's evaluated successes? How do you know the most efficient car to buy (government standardized test)? Most efficient refrigerator (government standardized test)? Most efficient hot water heater (government standardized test)? Or would you rather leave all those details in the hands of the lawyer-funded Consumer Reports or something?
It's blatantly obvious that higher efficiency equates to lower costs, so if the data centers don't care now, why will they care after the study is finished?
Data center operators do not know what an efficient system is, just as you probably do not know how your house's energy use compares to similar houses in your climate (unless you utilized the government's "useless" Energy Star database that has led to no regulation).
Here's a clue: If you live in a first world country there's a 100% chance your economy is based directly on the laissez-faire model. Chances are also high that the areas the most screwed up (i.e. telecom monopolies, etc) are the areas with the most government regulation and meddling. Until you can point out a single thriving economy based on socialism, communism, or whatever it is you're advocating, please STFU about how laissez-faire doesn't work.
You express scorn for publicly funded research (what this entire thread is about) and call it worthless, "socialist," especially if it does not lead to regulation (odd emphasis for you to make, but so be it). Thriving economies that fall under your distorted definition of socialist (ie, government funding of basic research, the subject of this thread) include: The US, United Kingdom, Ireland, Australia, Israel, Germany, France (OK, a bit of stretch there) - actually, can you give me a single country that DOES NOT fund basic research and provide data to the public that has a successful industry? Your absolutist vision of Laissez-faire does not work and has never been implemented (I would argue that based on his writings, Adam Smith clearly accepted the value of public investment in the commons and the extension of that to include intellectual development).
Intel is actually making a good push at moving the market to more efficient power supplies. That is going to happen. Data centers can be benchmarked, it's not simple but over the last few years we've developed some good metrics. Of course, prescriptive codes will probably never be applied - even California has not applied it's prescriptive codes to industrial spaces in the 30 odd years they been around. And if efficiency runs counter to reliability you need to fire your engineer. I know that does not have to be the case. And I'm appalled that you suggest that the cooling provided by the tiny air mass in a data center is anything like a usable buffer in case of a loss of cooling. The difference between 5 seconds to rack meltdown and 15 seconds to rack meltdown is irrelevant and I'd fire (or at least give a good educational berating) to any engineer who honestly suggested otherwise.
Do you have any idea how valuable the Energy Star program has been to just about every industry it touches? Your cheap shot makes you look stupid to anyone familiar with the documented reality of its impact. Improving the information distribution in a market is the best way to maximize market efficiency and having an independent entity (ideally the same as the contract enforcer, to use Adam Smith's terminology) provide the information format and dissemination is the cheapest way to do this.
I have seen more than once a big 'ol tank of water more than deal with long restart time issues (which I am not aware of with modern chillers, but could certainly exist if you're using crap Carriers or something in your plant). And payback is an awful way to assess the value of any measure - IRR or ROI calcs are the metrics businesses tend to prefer.
The most efficient data center design I'm working on at the moment is simpler to maintain than a standard system. And almost anything is easier than laying out five dozen individual CRAC units per building (miles of refrigerant piping, 120 compressors, 240 condenser fans) yet people still use that insane approach.
How is the EPA fast-tracking changes, or promoting ideas? I'm quite baffled how providing a database of typical data center power usage does anything like what you are suggesting, and they have certainly never used the similar databases they have created for other building types to do anything like that.
Your comment on taxes is moronic. Please clarify if you actually had an intelligent point.
Yes. I am saying data center efficiency is more important to society at large than to individual operators. This is a blatantly obvious fact and governments who have embraced it (for example, through the negawatt approach of avoiding capital-intensive new power plant projects used astonishingly successfully by many utility districts) have saved money and improved their environment. But don't let the well documented experiences of the past three decades get in the way of your heart felt (and never proven successful, anywhere, ever) libertarian/anarchist economic ideology.
Your clever idea to keep government out of doing research for business is one highly cherished by many third world countries. Not a single first world country is stupid enough to disregard decades of evidence and example to the contrary. I wonder why.
The government is meddling by providing free information to the market. This is as "meddlesome" as providing a standardized mpg rating or requiring nutritional information on food in the supermarket.
Other than an ideology as scorned by historical reality as Marxism, do you have any real-world examples of where your completely laissez-faire economy has ever worked? Even once?
Because your concerns about mercury are incredibly well documented to be absolutely stupid, since the mercury emissions to the environment from the coal based power grid in the US are far more harmful. Incandescent bulbs are a greater mercury pollution risk due to the coal plant emissions they directly require to operate (versus a comparable CFL), which ends up in and on the homes of at least half the US population, than the trivial amounts used in CFLs, no matter what the nice lady trying to get you to sign the petition to ban di-hydrogen monoxide told you. This is well documented and measured fact if you care about reality. And none of that "my power comes from clean hydro/nuclear/wind/fairy farts bullshit - it's a national grid (rickity, but functional) and a commodity product.
The free market is useless if true costs are not enforced by an outside authority on products. Energy waste is a classic tragedy of the commons to put it simply (since I don't want to waste my time getting into details of split incentives and the current regulatory climate's favoring of capital intensive projects over load reduction with you).
No real comment on daylight savings. I don't get my panties in a bunch about moving my clock by an hour. Other than a slightly batty neighbor-of-a-friend in Indiana, you're the only other person I've found who could give a rats ass about it.
I am an AC engineer, and it makes a hell of a lot of sense to have a standardized air outlet. But cooling densities are so high that airflow requirements dictate the outlet be about the size of the entire back of the rack...
And the government DOES NOT have to set a requirement, just provide a rational and well thought out specification. It is appropriate to allow the market to dictate adoption or rejection of the spec.
From wikipedia (although you can drag it from the original bill if you want to): "... when the federal Clean Energy Act of 2007[5] was signed into law on December 19, 2007. This legislation effectively banned (by January 2014) incandescent bulbs that produce 310 - 2600 lumens of light. Bulbs outside this range (roughly, light bulbs currently less than 40 Watts or more than 150 Watts) are exempt from the ban. Also exempt are several classes of specialty lights, including appliance lamps, "rough service" bulbs, 3-way, colored lamps, and plant lights."
Lets just say I'm not impressed when all your catastrophic examples (OMG! I won't be able to run my lava lamp or EZ Bake oven!!11!) are, in fact, completely made up. Not to mention completely irrelevant; regulating individual appliances is far different than regulating buildings, which the feds have never done despite having the data for decades. Why should I waste time responding to any of your other opinions or "reasoning?"
Sorry, but I oppose self-esteem coddling: What you wrote was stupid and you should be a bit embarrassed for having said it.
None of these elements are being destroyed, just shuffled about. Someday there may be quite a market for mining old landfills, or services that mine copper piping from houses and replace it with (bio-based) polymer piping. While an interesting article, I'm not too worried.
Sigh. Again, why hasn't the EPA regulated any of the dozen or so building types that have been in their Energy Star program for over a decade? Buildings that combined use far more energy than data centers? What would be the possible framework for such a federal mandate? Even California doesn't regulate data centers, and they have the data and have been regulating other building types since the 70's (enormously successfully BTW, leading to the most efficient building stock in the nation and absurdly low kW/capita requirements even after weather is taken into account).
The datacenter industry is highly competetive. While there are groups like the Uptime Institute that tackle large scale coordination, it is entirely in the government and society's best interest to offer public organization aid. And the taxes data centers pay MORE than covers the cost (and those taxes will NOT be coming in if they all move to India, a real risk as bandwidth becomes far cheaper than labor).
I am a consultant who works for and has worked for many large data center owners. I guarantee you that data center operators are not "smart" enough to figure out efficiency by them selves (hell, some aren't even smart enough to hire me;). Well, they are, but only after marketing, maintenance contracting, staffing, connectivity contracts, capacity contracts, generator testing, airflow balance for the new racks, battery service, diesel contract, annual maintenance scheduling, roof inspection, etc. is covered. There are still hundreds of large data centers (over 2 MW critical load) running with rows of air cooled computer room air conditioner (Liebert). This is breathtakingly inefficient in 95% of climates and maintenance intensive to boot - but they're still being built. Not to mention all the little guys; are you're saying that every business out there throwing a little 1000 sf data center in the basement of their new building can make a good business case to invest in all this research?
I think we have a fundamental disagreement on the value of government research, which has wasted money on such things as polio vaccines or integrated circuits, versus a pure free market, as they have in Somalia today.
The flaw in your reasoning is that I have found over the last dozen or so data centers I've looked at, is that operators have no idea if their data center is inefficient or not. None. They can't even tell if their old Lieberts are less efficient than their new Emerson Liebert units (you'd be surprised), or how their cooling tower setpoint impacts their plant kW/ton. No clue. Never cared.
The typical "cookie" I've seen offered to sites is information - the data on how their site is performing and the (anonymized) data on how that compares to everyone else's site. Smart operators realize it is only to their benefit if they can find out their UPS system is a dog wasting an extra half-million dollar a year, or their air-cooled CRAC system is sucking down double the power of everyone else's chiller plant.
I am somewhat dismayed at how many folks involved in Information Technology appear to be just wire pluggers who have no comprehension of how valuable information is.
Without regard to the sort of content being dealt whith, it's completely irrational to apply any sort of standards to data-center power consumption. A data-center that simply regurgitates static continent is going to have significantly different power requirements than a site that is actively dealing with processing and transcoding user-generated content. Compare Youtube to a high-volume brochure-ware site for an extreme example.
The best the EPA could produce, without creating an unnecessary burden on hosts would be to publish standards for OSes and hardware when idle. Anything beyond that is infeasible - govt mandated requirements to use O(log(n)) algorithms instead of O(n^2) ones would be patently absurd.
As a first point, the EPA is NOT looking to regulate data centers, just as they have not regulated schools, offices, hospitals, hotels, warehouses or any of the other types of buildings currently in the energy star database for the last decade or two.
Secondly, typically 37-50% of a data center's power usage goes to support systems: the cooling, UPS, and humidity control. Using the common metric PUE (Power Usage Effectiveness, or the SI-EER if you want to use the Uptime Institute's less smelly nomenclature) you can get a rough handle of the efficiency of a data center on this mechanical component - Total Data Center Power / UPS Output Power.
Completely off the table at the moment are the savings possible from virtualization and implementation of on-board power schemes (sleep modes and such). There is no good standardized way to measure this yet due to the types of problems you mention. But it is exceptionally easy to measure the amount of critical load in a data center (UPS output watts - done), although even that has some unexpected blurry edges (small variable speed fans within the rack can have surprisingly large kW swings in response to different mechanical fan schemes).
And again, why do you think the EPA is going to suddenly want to regulate data centers when they have never shown any inclination over the last couple decades to regulate any other type of building?
Question: How has the government used the enormous database of commercial building power usage, which is still larger in aggregate than data centers? Other than providing a useful tool for setting design goals, what nefarious federal regulations have been spawned by this evil data collection, analysis, and presentation?
You're a little optimistic about data center's approach to their own best interest. There has been quite a bit of work done in the area of data center best practices and not a heck of a lot of interest. For example, hot aisle / cold aisle is still honestly debated and considered a 'new' idea in many legacy data centers.
Power is still quite cheap and very low on the site totem pole below connectivity and reliability.
The government acting as a neutral party to collect useful data is hardly meddling. Next you'll decide to rant about them meddling in pharmaceuticals by funding cancer research.
The facts of the matter are that datacenter energy use is very poorly understood by owners and considered a negligible cost of the business. When benchmarked, datacenter efficiency has been found to vary by over a factor of two between facilities. Owners don't really know what is efficient - high bills are just part of the business, and competitors aren't willing to share good data on the subject. Hence the need for the government to provide some benchmarks, similar to the flawed, but better than nothing, mpg ratings for cars.
The government, with access to regulated utility records, knows exactly how much power a datacenter is using. And the energy star program applied for years to other commercial buildings has resulted in no federal mandates. You're basically pushing a line of FUD for... well no one benefits really. Hopefully you at least get a good Funny mod rating for your efforts to stunt the development of useful data to help the industry.
While I think desert datacenters are a bad plan, dust is an easily solved issue. Filters are an incredibly mature technology, and so little outside air is brought in (basically just enough to keep the floor positively pressurized so nothing sneaks in through the cracks) it's a non-issue.
That said, the most efficient datacenter designs I see use 100% outside air (with appropriate low-face velocity, low fan power filtration) much of the time to cool the space, but their in temperate climes.
Most big datacenters I've seen are cited to a great extent based on the availability of power. Finding 20+ MW of unused capacity of adequate reliability is difficult, and it is expensive to have that scale infrastructure built out just for your datacenter. I have heard of datacenters catching the 'green' bug when they ran out of power and were told tough - build your own damn plant then. The other issue is of course good feeds to the internet, although that seems to be coming up less and less as a problem.
Speaking strictly to the cooling generation side of things, the biggest thing that saves energy is implementing freecooling, that is bringing in outside air directly when it is cold and using it to cool the building (contamination is a easy known problem to deal with - filtering is not hard). If you're in a dry climate, use a cooling tower to make cold water and use that in coils. Blindingly simple, but datacenters just don't do it, even though their 24/7 load that is independent of outdoor air temperature is a great match for it. Part of the reason is it doesn't save peak load, and peak kW is where it's really at. Many datacenters I've seen have maxed out their utility feed, and paying for new infrastructure on the MW scale is not cheap.
I agree on code, but (as a ME efficiency weanie) hardware is obsolete and should be replaced when it become unsustainably inefficient. For example, if you can buy an equivalent replacement unit, deploy it, and pay to power it for a year with the cost of just one year's kWh bill on the legacy box, that old hareware is obsolete and should be tossed.
And most big datacenters tend to have at least a few dozen kW (sometimes over a hundred) of such zombie servers, sucking down power 24/7 for one or two batch process a month that could be virtualized on my laptop in an emulator run in Windows for god's sake... Seriously, in one audit session it was proposed to create a zombie bounty to reward IT staff for identifying and bagging the easy money savings (and capacity increase for watt-hungry sites) from killing old hardware.
Providing electricity to parking areas in apartment buildings is not much more difficult or costly than providing lighting. In cold climates, parking slots with 120V plug ins are routinely provided for apartment dwellers to power block heaters (for example, check the specs on Minot University student housing apartments in North Dakota). Hell, I have heard that sometimes even movie theaters have them for public use in non-assigned parking spaces (any Alaska dwellers with first-person support?). The only reason that there is a 'last 10 feet' problem is that there is absolutely bone-zero real demand for a solution. The electrical infrastructure is really a non-issue, it's the rolling storage that's the hold-up.
As TinyManCan points out, there is plenty of capacity available (albeit much of it is nuke or coal) to charge cars offpeak. For example, if you watch California's demand curve for a while, http://www.caiso.com/outlook/SystemStatus.html, you'll notice that the grid is at only about half capacity at night. There is less, but still significant, surplus available in the morning (post commute). Trying to follow this wonky demand curve with coal fired plants that throttle back slower than the Titanic is the ongoing challenge of every utility company. Grid-intertied electric vehicles, which give back a kW or two for short period during the peak hour, are actively being investigated to *help* with the grid capacity problems. PG&E is actually pursuing buying 'dead,' ie less than 50% design capacity, hybrid batteries to hookup to the grid to add a bit of cheap storage to help absorb spikes.
It's easy to throw up good sounding, but embarrassingly and utterly wrong, arguments against reasonable enviromental ideas that have been well understood and vetted for over three decades (awaiting only a good battery really - the Prius-spawned supply chain has matured the necessary power electronics and motor package nicely). I know there aren't suppose to be any stupid questions, but asking, "But where are we going to get the power to charge them, huh huh?" is similar to arguing that a mission to the moon is impossible not because of the daunting energy requirements and orbital mechanics but rather because "It's obviously impossible for green cheese to support the weight of the lander."
Slashdot Mirror
The real benefit to a 100F setpoint is the free cooling it allows. You can use filtered out door air or evaporatively cooled water from a cooling tower to keep a datacenter at 100F year round just about anywhere. This is a 90% reduction in cooling energy right there using decades old, mature HVAC tech.
The main benefit is that it allows free cooling, either bringing in filtered out door air directly or using a cooling tower to generate evaporatively cooled water for direct use in coils. At a space temperature of 100F, you could cut out 90% of the cooling - even in a full-time humid climate like Malaysia. In most US climes, you can do better.
I'm sorry, but you still haven't shown any proof that this study needed to be done by the government. You can try to change the subject all you want, but that doesn't change the fact that any competitive data center would be aiming for higher efficiency because it directly lowers their costs and allows them to offer lower prices than their competitors.
This just is not true. I work in the industry, performing studies of this nature. The Uptime Institute, 24/7 Group, etc. are not enough to promote sharing of this data between rivals. In a similar arena (semiconductor cleanroom critical environments), I've been paid to do the same damn study three times - but the third and last time it was funded by the government (LBNL) and disseminated publicly (something industry groups, such as SEMATECH, do not do). That is efficiency that saves money on an industry wide scale (and I have plenty of work, I don't need to do the same thing over and over for every little group of companies out there). The government's investment saved industry ten, possibly a hundred fold, expenditure. That's good business on a national economy scale. Who else would be willing to invest in the economic commons?
Besides that, if the government isn't going to pass legislation based on the study results, why even bother?
The Energy Star program has proven across numerous fields the value of providing information to the market. Are you at all familiar with it's evaluated successes? How do you know the most efficient car to buy (government standardized test)? Most efficient refrigerator (government standardized test)? Most efficient hot water heater (government standardized test)? Or would you rather leave all those details in the hands of the lawyer-funded Consumer Reports or something?
It's blatantly obvious that higher efficiency equates to lower costs, so if the data centers don't care now, why will they care after the study is finished?
Data center operators do not know what an efficient system is, just as you probably do not know how your house's energy use compares to similar houses in your climate (unless you utilized the government's "useless" Energy Star database that has led to no regulation).
Here's a clue: If you live in a first world country there's a 100% chance your economy is based directly on the laissez-faire model. Chances are also high that the areas the most screwed up (i.e. telecom monopolies, etc) are the areas with the most government regulation and meddling. Until you can point out a single thriving economy based on socialism, communism, or whatever it is you're advocating, please STFU about how laissez-faire doesn't work.
You express scorn for publicly funded research (what this entire thread is about) and call it worthless, "socialist," especially if it does not lead to regulation (odd emphasis for you to make, but so be it). Thriving economies that fall under your distorted definition of socialist (ie, government funding of basic research, the subject of this thread) include: The US, United Kingdom, Ireland, Australia, Israel, Germany, France (OK, a bit of stretch there) - actually, can you give me a single country that DOES NOT fund basic research and provide data to the public that has a successful industry? Your absolutist vision of Laissez-faire does not work and has never been implemented (I would argue that based on his writings, Adam Smith clearly accepted the value of public investment in the commons and the extension of that to include intellectual development).
Intel is actually making a good push at moving the market to more efficient power supplies. That is going to happen. Data centers can be benchmarked, it's not simple but over the last few years we've developed some good metrics. Of course, prescriptive codes will probably never be applied - even California has not applied it's prescriptive codes to industrial spaces in the 30 odd years they been around. And if efficiency runs counter to reliability you need to fire your engineer. I know that does not have to be the case. And I'm appalled that you suggest that the cooling provided by the tiny air mass in a data center is anything like a usable buffer in case of a loss of cooling. The difference between 5 seconds to rack meltdown and 15 seconds to rack meltdown is irrelevant and I'd fire (or at least give a good educational berating) to any engineer who honestly suggested otherwise.
Do you have any idea how valuable the Energy Star program has been to just about every industry it touches? Your cheap shot makes you look stupid to anyone familiar with the documented reality of its impact. Improving the information distribution in a market is the best way to maximize market efficiency and having an independent entity (ideally the same as the contract enforcer, to use Adam Smith's terminology) provide the information format and dissemination is the cheapest way to do this.
I have seen more than once a big 'ol tank of water more than deal with long restart time issues (which I am not aware of with modern chillers, but could certainly exist if you're using crap Carriers or something in your plant). And payback is an awful way to assess the value of any measure - IRR or ROI calcs are the metrics businesses tend to prefer.
The most efficient data center design I'm working on at the moment is simpler to maintain than a standard system. And almost anything is easier than laying out five dozen individual CRAC units per building (miles of refrigerant piping, 120 compressors, 240 condenser fans) yet people still use that insane approach.
How is the EPA fast-tracking changes, or promoting ideas? I'm quite baffled how providing a database of typical data center power usage does anything like what you are suggesting, and they have certainly never used the similar databases they have created for other building types to do anything like that.
Your comment on taxes is moronic. Please clarify if you actually had an intelligent point.
Yes. I am saying data center efficiency is more important to society at large than to individual operators. This is a blatantly obvious fact and governments who have embraced it (for example, through the negawatt approach of avoiding capital-intensive new power plant projects used astonishingly successfully by many utility districts) have saved money and improved their environment. But don't let the well documented experiences of the past three decades get in the way of your heart felt (and never proven successful, anywhere, ever) libertarian/anarchist economic ideology.
Your clever idea to keep government out of doing research for business is one highly cherished by many third world countries. Not a single first world country is stupid enough to disregard decades of evidence and example to the contrary. I wonder why.
The government is meddling by providing free information to the market. This is as "meddlesome" as providing a standardized mpg rating or requiring nutritional information on food in the supermarket.
Other than an ideology as scorned by historical reality as Marxism, do you have any real-world examples of where your completely laissez-faire economy has ever worked? Even once?
Because your concerns about mercury are incredibly well documented to be absolutely stupid, since the mercury emissions to the environment from the coal based power grid in the US are far more harmful. Incandescent bulbs are a greater mercury pollution risk due to the coal plant emissions they directly require to operate (versus a comparable CFL), which ends up in and on the homes of at least half the US population, than the trivial amounts used in CFLs, no matter what the nice lady trying to get you to sign the petition to ban di-hydrogen monoxide told you. This is well documented and measured fact if you care about reality. And none of that "my power comes from clean hydro/nuclear/wind/fairy farts bullshit - it's a national grid (rickity, but functional) and a commodity product. The free market is useless if true costs are not enforced by an outside authority on products. Energy waste is a classic tragedy of the commons to put it simply (since I don't want to waste my time getting into details of split incentives and the current regulatory climate's favoring of capital intensive projects over load reduction with you). No real comment on daylight savings. I don't get my panties in a bunch about moving my clock by an hour. Other than a slightly batty neighbor-of-a-friend in Indiana, you're the only other person I've found who could give a rats ass about it.
I am an AC engineer, and it makes a hell of a lot of sense to have a standardized air outlet. But cooling densities are so high that airflow requirements dictate the outlet be about the size of the entire back of the rack...
And the government DOES NOT have to set a requirement, just provide a rational and well thought out specification. It is appropriate to allow the market to dictate adoption or rejection of the spec.
From wikipedia (although you can drag it from the original bill if you want to): "... when the federal Clean Energy Act of 2007[5] was signed into law on December 19, 2007. This legislation effectively banned (by January 2014) incandescent bulbs that produce 310 - 2600 lumens of light. Bulbs outside this range (roughly, light bulbs currently less than 40 Watts or more than 150 Watts) are exempt from the ban. Also exempt are several classes of specialty lights, including appliance lamps, "rough service" bulbs, 3-way, colored lamps, and plant lights."
Lets just say I'm not impressed when all your catastrophic examples (OMG! I won't be able to run my lava lamp or EZ Bake oven!!11!) are, in fact, completely made up. Not to mention completely irrelevant; regulating individual appliances is far different than regulating buildings, which the feds have never done despite having the data for decades. Why should I waste time responding to any of your other opinions or "reasoning?"
Sorry, but I oppose self-esteem coddling: What you wrote was stupid and you should be a bit embarrassed for having said it.
None of these elements are being destroyed, just shuffled about. Someday there may be quite a market for mining old landfills, or services that mine copper piping from houses and replace it with (bio-based) polymer piping. While an interesting article, I'm not too worried.
Sigh. Again, why hasn't the EPA regulated any of the dozen or so building types that have been in their Energy Star program for over a decade? Buildings that combined use far more energy than data centers? What would be the possible framework for such a federal mandate? Even California doesn't regulate data centers, and they have the data and have been regulating other building types since the 70's (enormously successfully BTW, leading to the most efficient building stock in the nation and absurdly low kW/capita requirements even after weather is taken into account).
The datacenter industry is highly competetive. While there are groups like the Uptime Institute that tackle large scale coordination, it is entirely in the government and society's best interest to offer public organization aid. And the taxes data centers pay MORE than covers the cost (and those taxes will NOT be coming in if they all move to India, a real risk as bandwidth becomes far cheaper than labor).
I am a consultant who works for and has worked for many large data center owners. I guarantee you that data center operators are not "smart" enough to figure out efficiency by them selves (hell, some aren't even smart enough to hire me ;). Well, they are, but only after marketing, maintenance contracting, staffing, connectivity contracts, capacity contracts, generator testing, airflow balance for the new racks, battery service, diesel contract, annual maintenance scheduling, roof inspection, etc. is covered. There are still hundreds of large data centers (over 2 MW critical load) running with rows of air cooled computer room air conditioner (Liebert). This is breathtakingly inefficient in 95% of climates and maintenance intensive to boot - but they're still being built. Not to mention all the little guys; are you're saying that every business out there throwing a little 1000 sf data center in the basement of their new building can make a good business case to invest in all this research?
I think we have a fundamental disagreement on the value of government research, which has wasted money on such things as polio vaccines or integrated circuits, versus a pure free market, as they have in Somalia today.
The flaw in your reasoning is that I have found over the last dozen or so data centers I've looked at, is that operators have no idea if their data center is inefficient or not. None. They can't even tell if their old Lieberts are less efficient than their new Emerson Liebert units (you'd be surprised), or how their cooling tower setpoint impacts their plant kW/ton. No clue. Never cared.
The typical "cookie" I've seen offered to sites is information - the data on how their site is performing and the (anonymized) data on how that compares to everyone else's site. Smart operators realize it is only to their benefit if they can find out their UPS system is a dog wasting an extra half-million dollar a year, or their air-cooled CRAC system is sucking down double the power of everyone else's chiller plant.
I am somewhat dismayed at how many folks involved in Information Technology appear to be just wire pluggers who have no comprehension of how valuable information is.
Without regard to the sort of content being dealt whith, it's completely irrational to apply any sort of standards to data-center power consumption. A data-center that simply regurgitates static continent is going to have significantly different power requirements than a site that is actively dealing with processing and transcoding user-generated content. Compare Youtube to a high-volume brochure-ware site for an extreme example.
The best the EPA could produce, without creating an unnecessary burden on hosts would be to publish standards for OSes and hardware when idle. Anything beyond that is infeasible - govt mandated requirements to use O(log(n)) algorithms instead of O(n^2) ones would be patently absurd.
As a first point, the EPA is NOT looking to regulate data centers, just as they have not regulated schools, offices, hospitals, hotels, warehouses or any of the other types of buildings currently in the energy star database for the last decade or two.
Secondly, typically 37-50% of a data center's power usage goes to support systems: the cooling, UPS, and humidity control. Using the common metric PUE (Power Usage Effectiveness, or the SI-EER if you want to use the Uptime Institute's less smelly nomenclature) you can get a rough handle of the efficiency of a data center on this mechanical component - Total Data Center Power / UPS Output Power.
Completely off the table at the moment are the savings possible from virtualization and implementation of on-board power schemes (sleep modes and such). There is no good standardized way to measure this yet due to the types of problems you mention. But it is exceptionally easy to measure the amount of critical load in a data center (UPS output watts - done), although even that has some unexpected blurry edges (small variable speed fans within the rack can have surprisingly large kW swings in response to different mechanical fan schemes).
And again, why do you think the EPA is going to suddenly want to regulate data centers when they have never shown any inclination over the last couple decades to regulate any other type of building?
Question: How has the government used the enormous database of commercial building power usage, which is still larger in aggregate than data centers? Other than providing a useful tool for setting design goals, what nefarious federal regulations have been spawned by this evil data collection, analysis, and presentation?
You're a little optimistic about data center's approach to their own best interest. There has been quite a bit of work done in the area of data center best practices and not a heck of a lot of interest. For example, hot aisle / cold aisle is still honestly debated and considered a 'new' idea in many legacy data centers. Power is still quite cheap and very low on the site totem pole below connectivity and reliability.
The government acting as a neutral party to collect useful data is hardly meddling. Next you'll decide to rant about them meddling in pharmaceuticals by funding cancer research. The facts of the matter are that datacenter energy use is very poorly understood by owners and considered a negligible cost of the business. When benchmarked, datacenter efficiency has been found to vary by over a factor of two between facilities. Owners don't really know what is efficient - high bills are just part of the business, and competitors aren't willing to share good data on the subject. Hence the need for the government to provide some benchmarks, similar to the flawed, but better than nothing, mpg ratings for cars.
The government, with access to regulated utility records, knows exactly how much power a datacenter is using. And the energy star program applied for years to other commercial buildings has resulted in no federal mandates. You're basically pushing a line of FUD for... well no one benefits really. Hopefully you at least get a good Funny mod rating for your efforts to stunt the development of useful data to help the industry.
While I think desert datacenters are a bad plan, dust is an easily solved issue. Filters are an incredibly mature technology, and so little outside air is brought in (basically just enough to keep the floor positively pressurized so nothing sneaks in through the cracks) it's a non-issue. That said, the most efficient datacenter designs I see use 100% outside air (with appropriate low-face velocity, low fan power filtration) much of the time to cool the space, but their in temperate climes.
Most big datacenters I've seen are cited to a great extent based on the availability of power. Finding 20+ MW of unused capacity of adequate reliability is difficult, and it is expensive to have that scale infrastructure built out just for your datacenter. I have heard of datacenters catching the 'green' bug when they ran out of power and were told tough - build your own damn plant then. The other issue is of course good feeds to the internet, although that seems to be coming up less and less as a problem.
Speaking strictly to the cooling generation side of things, the biggest thing that saves energy is implementing freecooling, that is bringing in outside air directly when it is cold and using it to cool the building (contamination is a easy known problem to deal with - filtering is not hard). If you're in a dry climate, use a cooling tower to make cold water and use that in coils. Blindingly simple, but datacenters just don't do it, even though their 24/7 load that is independent of outdoor air temperature is a great match for it. Part of the reason is it doesn't save peak load, and peak kW is where it's really at. Many datacenters I've seen have maxed out their utility feed, and paying for new infrastructure on the MW scale is not cheap.
I agree on code, but (as a ME efficiency weanie) hardware is obsolete and should be replaced when it become unsustainably inefficient. For example, if you can buy an equivalent replacement unit, deploy it, and pay to power it for a year with the cost of just one year's kWh bill on the legacy box, that old hareware is obsolete and should be tossed. And most big datacenters tend to have at least a few dozen kW (sometimes over a hundred) of such zombie servers, sucking down power 24/7 for one or two batch process a month that could be virtualized on my laptop in an emulator run in Windows for god's sake... Seriously, in one audit session it was proposed to create a zombie bounty to reward IT staff for identifying and bagging the easy money savings (and capacity increase for watt-hungry sites) from killing old hardware.
Providing electricity to parking areas in apartment buildings is not much more difficult or costly than providing lighting. In cold climates, parking slots with 120V plug ins are routinely provided for apartment dwellers to power block heaters (for example, check the specs on Minot University student housing apartments in North Dakota). Hell, I have heard that sometimes even movie theaters have them for public use in non-assigned parking spaces (any Alaska dwellers with first-person support?). The only reason that there is a 'last 10 feet' problem is that there is absolutely bone-zero real demand for a solution. The electrical infrastructure is really a non-issue, it's the rolling storage that's the hold-up.
It's easy to throw up good sounding, but embarrassingly and utterly wrong, arguments against reasonable enviromental ideas that have been well understood and vetted for over three decades (awaiting only a good battery really - the Prius-spawned supply chain has matured the necessary power electronics and motor package nicely). I know there aren't suppose to be any stupid questions, but asking, "But where are we going to get the power to charge them, huh huh?" is similar to arguing that a mission to the moon is impossible not because of the daunting energy requirements and orbital mechanics but rather because "It's obviously impossible for green cheese to support the weight of the lander."