Microsoft, Meta and Others Face Rising Drought Risk to Their Data Centers (cnbc.com) 69
"Drought conditions are worsening in the U.S.," reports CNBC, "and that is having an outsized impact on the real estate that houses the internet."
Water is the cheapest and most common method used to cool the centers. In just one day, the average data center could use 300,000 gallons of water to cool itself — the same water consumption as 100,000 homes, according to researchers at Virginia Tech who also estimated that one in five data centers draws water from stressed watersheds mostly in the west. "There is, without a doubt, risk if you're dependent on water," said Kyle Myers, vice president of environmental health, safety & sustainability at CyrusOne, which owns and operates over 40 data centers in North America, Europe, and South America. "These data centers are set up to operate 20 years, so what is it going to look like in 2040 here, right...?"
Realizing the water risk in New Mexico, Meta, formerly known as Facebook, ran a pilot program on its Los Lunas data center to reduce relative humidity from 20% to 13%, lowering water consumption. It has since implemented this in all of its center. But Meta's overall water consumption is still rising steadily, with one fifth of that water last year coming from areas deemed to have "water stress," according to its website. It does actively restore water and set a goal last year to restore more water than it consumes by 2030, starting in the west.
Microsoft has also set a goal to be "water positive" by 2030. Â"The good news is we've been investing for years in ongoing innovation in this space so that fundamentally we can recycle almost all of the water we use in our data centers," said Brad Smith, president of Microsoft. "In places where it rains, like the Pacific Northwest where we're headquartered in Seattle, we collect rain from the roof. In places where it doesn't rain like Arizona, we develop condensation techniques."
Realizing the water risk in New Mexico, Meta, formerly known as Facebook, ran a pilot program on its Los Lunas data center to reduce relative humidity from 20% to 13%, lowering water consumption. It has since implemented this in all of its center. But Meta's overall water consumption is still rising steadily, with one fifth of that water last year coming from areas deemed to have "water stress," according to its website. It does actively restore water and set a goal last year to restore more water than it consumes by 2030, starting in the west.
Microsoft has also set a goal to be "water positive" by 2030. Â"The good news is we've been investing for years in ongoing innovation in this space so that fundamentally we can recycle almost all of the water we use in our data centers," said Brad Smith, president of Microsoft. "In places where it rains, like the Pacific Northwest where we're headquartered in Seattle, we collect rain from the roof. In places where it doesn't rain like Arizona, we develop condensation techniques."
Dry Sump (Score:5, Insightful)
Re: Dry Sump (Score:5, Interesting)
Re: Dry Sump (Score:5, Interesting)
Heated up and evaporated most likely. You really can't just dump hot water into a river anymore. Even if they did, the water is not lost.
Going up the cooling tower though, that rains back out far away.
300,000 gallons is as much as 100,000 homes? Three gallons a day per household?
Someone needs to check their own water bill.
Re: (Score:3)
Re: (Score:2)
Re:Dry Sump (Score:5, Informative)
My car uses a dry sump oil system. Why can a data center not use a similar system to cool and constantly reuse that water?
What a dry sump oil system means is that there's a scavenging pump at the bottom instead of a sump, and the oil is whisked off to a reservoir instead of sitting beneath it. The purpose is to prevent oil sloshing around in the sump from either not being picked up properly or getting in the way of the rotating assembly and being churned up which can cause foaming (and which also costs power as the oil is smacked around) and it has nothing to do whatsoever with this. Your bragging about your fancy engine is irrelevant to this issue.
With that said, the reason they don't use cooling loops for most datacenters is that you have to sink the heat somewhere, and most of the threatened DCs are in hot climates where you would need a huge amount of heat exchanger area to even get it cooled down to near ambient, which would still be much warmer than the fresh water supply.
Re: Dry Sump (Score:2)
Re: (Score:2)
In sports cars, they are also used to displace the heat in the oil as the oil can get quite hot
Oil heat is removed from the oil by an oil cooler whether the sump is wet or dry. Having a remote sump does make it easier to keep the oil cooler, because less engine heat is transferred into the oil since it's not sitting in the sump, but a slightly bigger oil cooler will also do that job. The water used for coolant in the DCs isn't sitting around in a hot server after it's been used for cooling, so there's no parallel there regardless.
Re: (Score:2)
Re: (Score:2)
They are very rare in production autos, but there are a few, like the Corvette Z06. It also lets you mount the engine lower, due to the reduced sump, so it's really very beneficial. It's usually only really used in racing.
Re: (Score:1)
Thermodynamic positive (Score:1)
Re:Thermodynamic positive (Score:5, Funny)
Water positive.... I'm glad MS had made the goal to produce more water than it uses. Since it has the ability already to magically produce more water than they use why don't they just do that?
I guess they will buy hydrogen, burn it, and collect the product. Better still, they will make it from protons and neutrons they get from somewhere.
Re: (Score:2)
I think the idea is to condensate from the air and return it to the ground water supplies for others to use. Needs lot of energy but when that comes from renewables then it works out.
Re: Thermodynamic positive (Score:2)
They seem to trying a few things, such as collecting and processing rainwater. Condensing water from the air is pretty inefficient and unlikely to work well in places with stressed supply. It works best in humid areas, those areas tending to also be ones with the most water to begin with.
Alto their credit, they seem to be aiming to get water into areas most stressed by short supply. My initial thought was that this'd a pointless gesture akin to planting trees elsewhere as credit for deforestation.
https://bl [microsoft.com]
3 gallons per home per day?? Bullshit (Score:5, Interesting)
How many libraries of Congress is that?
Fucking stupid to say the average home uses only 3 gallons a day.
First google hit for "gallons per minute showering" says this,
> The average American shower uses approximately 15.8 gallons (59.8 liters) and lasts for 7.8 minutes at an average flow rate of 2.1 gallons per minute (7.9 lpm).
Would the editors please do some trivial fact checking before posting? Or maybe it's true that 2/3rds of homes don't shower daily and use no other water at all. *eye roll*
Yet another reason this place is dying.
Re:3 gallons per home per day?? Bullshit (Score:5, Informative)
People use more than 3 gallons per day just flushing toilets.
Re: (Score:2)
3 gallons per day just flushing toilets
I use twice that just flushing my illegal 6 gpf toilet once.
Re: (Score:2)
300,000 gallons is about $100 worth of water at commercial rates.
TFA is the stupidest thing I have read so far today.
Re: (Score:2)
Says the noob with the 8-digit UID.
People were making the same complaint back when I got my 5-digit UID, which was about a week after the DNS records propagated for slashdot.org
The people meant to do the fact-checking are the reporters writing the stories (or in Slashdot terms, the users, submitting their summaries of stories). When I worked with professional newspaper editors, they also exp
Re: (Score:1)
Hi kiddo, I was here reading when they had 3 digit UID available.
I was here when anon could post.
I was here when they killed that.
Dismissing someone because of their UID is... hmmm.. which word to use for this one today... let's go with... idiotic.
Do you have anything at all to say about the content of my post? No? Then go back to twitter where you belong.
Have a nice day, kid. Have a wonderful day!
Re: (Score:2)
Prove it, noob. Or are you going to bleat the bleat about having lost your old UID/pw combination (which translates into English as "I am incompetent and should not be allowed near computers without adult supervision).
As for criticising your content, try reading the bit about where reporters (Slashdot submitters, and commentators) are responsible for fact-checking their content, not editors. Or is that beyond your 3s attention span?
Re: (Score:2)
That's not even 1 Danzig
Too Bad (Score:5, Insightful)
Too Bad, they should have chosen better areas to build their data centers. Maybe in places that get real winter and have plenty of water. Instead they chose warm areas that are water poor.
For example, lets take Minnesota. They can open the windows and get plenty of cooling in the winter :) From there going east, there is plenty of water to be had too.
No sympathy from me.
Re:Too Bad (Score:5, Interesting)
Checking https://droughtmonitor.unl.edu... [unl.edu], you'll note that many areas of the east are currently under drought conditions. We'd better hope this is not the new normal or these water problems of the bit-barns will be the least of our worries.
Re: (Score:2)
Checking https://droughtmonitor.unl.edu... [unl.edu], you'll note that many areas of the east are currently under drought conditions. We'd better hope this is not the new normal or these water problems of the bit-barns will be the least of our worries.
We do get occasional droughts here in the east. They tend to be at the times we aren't under flooding conditions.
Re: (Score:2)
Most of them are in better places (Score:2)
Of course, you don't want them all in one place, or if the power grid goes down somewhere you don't have backup.
The environmental issue is probably more complicated than it looks. The math is obviously wrong on
Re: (Score:2)
on balance, data centers and cloud infrastructure are a big resource for having people work from home rather than having to commute, which maybe has more negative impact.
Very good chance. It takes about 1.5 gallons of water to process 1 gallon of crude [energy.gov]...
20 year data center life? Bullshit (Score:1)
Complete nonsense right from summary says, "These data centers are set up to operate 20 years, so what is it going to look like in 2040 here, right...?"
Wtf? Now data centers are like some sort of natural resource that expires or like nuclear power plants leaving radio active waste to clean up?
For those in other fields, a data center is a big building with nothing but power, air conditioning, lots of internet, a bunch of batteries & generators, and some rent-a-cops for security. Which part has a 20 year
Re: (Score:1)
You really that daft that you cannot make the connection between their water use and dry conditions in their areas, and then pose the question of what those areas will be like in 2040? Bit-barns do not automatically regenerate themselves.
Re: (Score:1)
I am perfectly aware. And so what? If a few data centers are converted to warehouse space what harm was there? By the 20 year mark all those batteries have been replaced a few times, the AC at least 2-3 times, the rent a cops from the opening have long since quit and found real jobs, and so on. Why does it matter if a bunch of data centers shut down in 20 years for lack of water?
The article in same context also says those data centers are everywhere. Are we going dry everywhere? Or did we target only
Re: (Score:2)
Wtf? Now data centers are like some sort of natural resource that expires or like nuclear power plants leaving radio active waste to clean up?
No one has said that except you. Practically all buildings have an expected operating lifetime. Buildings and infrastructure do not "expire" 1 day after that lifetime. Rather it is the amount of time that an owner can expect to operate the building without major costly repairs and upgrades. With the building designed to be a data center, it is not likely that it can be repurposed as a general office building so it has to last another 20 years as a data center.
For those in other fields, a data center is a big building with nothing but power, air conditioning, lots of internet, a bunch of batteries & generators, and some rent-a-cops for security. Which part has a 20 year life span requiring the data center be shut down and what's that have to do with water usage?
[sarcasm]Air conditioners never servicing and r
Re: (Score:2)
With the building designed to be a data center, it is not likely that it can be repurposed as a general office building
Why not? Without the hardware, most DCs are big empty rooms with raised floors and/or drop ceilings. You could certainly use them for office buildings. I think it would arguably make more sense to use them for something more industrial like some kind of high-tech light manufacturing, but they certainly could have some walls erected and be offices. It just wouldn't really take advantage of the characteristics of the building.
Re: (Score:2)
Re: (Score:2)
It seems like you are disagreeing with the "Now data centers are essentially disposable..?!?!" statement, but I'm not following your argument.
Are you saying they do or don't try to keep it going past 20 years?
Do they perform general and proactive maintenance to these facilities? Yet still they can only expect ~20 years? (pathetic)
(or) Do they scrape by with bare minimum, patch and pray style maintenance? (idiotic)
(or) Do they just let it go to shit over 20+ years? (pathetic and idiotic)
Re: (Score:2)
It seems like you are disagreeing with the "Now data centers are essentially disposable..?!?!" statement, but I'm not following your argument.
No. I am pointing out basic engineering and construction principles: A commercial building having a operating lifetime of 20 years is not unusual. His incredulity that a building has an operating lifetime underlies he knows jack shit about engineering and construction. Operating lifetime does not mean expiration date in that everything fails 1 day after the operating lifetime.
Are you saying they do or don't try to keep it going past 20 years?
I'm saying that basic engineering principles factor in an operating lifetime or life expectancy. Take for example industrial pumps ca [pumpsandsystems.com]
Re: (Score:2)
Thank you for that, I now understand completely. I figured that 'cheaply as possible' was the driving factor, and this premise answers everything following that.
My 'pathetic' comment was based on what I suspected in the first place. It is not to say that the engineers, and builders/maintainers are bad at their jobs. On the contrary they are clearly hamstrung, forced to compromise by choosing from the lesser of evils. Primarily due to budget constraints imposed by the fool in charge. (i.e., "up to the ow
Data centres in space? (Score:2)
Use solar panels for energy, perhaps with an additional small reactor.. Put the hardware in the shade of the solar panels. Energy + cooling in one package. Not very good for low latency computing, but okay for jobs where the run takes a longer time, and the results don't have to be immediately available.
(Probably won't work, or be cost effective. Just a speculative daydream.)
Best wishes
Bob
Re: (Score:2)
That solution lacks a sense of proportion. You won't cool those data centers by parking them in the shade.
Re: (Score:2)
From the NASA website.
"Without thermal controls, the temperature of the orbiting Space Station's Sun-facing side would soar to 250 degrees F (121 C), while thermometers on the dark side would plunge to minus 250 degrees F (-157 C)."
That's the sort of thing I had in mind.
Best wishes,
Bob
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Know what one of the hardest things to do in a place without an atmosphere is, even if that place is really cold? Get rid of waste heat.
I didn't know this. Thanks.
Best wishes,
Bob
Re: (Score:2)
It doesn't make sense to launch servers to space, for what it costs you could have built a cooling system here. And just imagine the service fees.
Numbers seem wrong (Score:5, Insightful)
300,000 gallons divided by 100,000 homes equals 3 gallons per day per home.
Does that seem a little low?
Those would be some mighty short showers...
Re: (Score:2)
First result on Google:
>The average American family uses more than 300 gallons of water per day at home. Roughly 70 percent of this use occurs indoors. In addition, there are other miscellaneous uses of water in the house which may be very significant, depending on the degree of water conservation by the household.
So 2 magnitudes off...
Re: (Score:2)
Re: (Score:2)
Yeah, totally bogus. Virginia Tech needs to go back to some basic research.
https://www.epa.gov/watersense... [epa.gov]
Re: (Score:2)
300,000 gallons divided by 100,000 homes equals 3 gallons per day per home.
Does that seem a little low?
Those would be some mighty short showers...
I guess the people in those 100,000 homes scrub thoroughly between drips of the shower head?
Water Use (Score:4, Interesting)
Water use is from evaporation in cooling towers, plus blow-down needed to reduce suspended solids in condenser water as water evaporates. There are technologies available to reduce blow-down some (which cities should mandate retroactively), but it has an energy penalty.
You can also design systems so that you only use evaporation mode cooling when outside temperatures are over ~100F, and use a dry-cooling mode the rest of the time. Both changes increase electricity use to reduce water use.
Beyond that, your best bet for an energy-neutral solution is to dry-cool at night and have a very large thermal storage system (water rather than ice) to cover the loads in the day. For a large data center that would be over 4,000 acre-feet or 5 million m^3 of water storage.
Re: (Score:2)
It's funny how that a week or two back, on Slashdot, there was a bunch of comments about how these locations were perfect for data centres exactly because they all are built as "water positive" already.
Re: (Score:2)
Ha, re-reading the comments I see I had too rosy an impression. They clearly are not yet there - https://slashdot.org/story/22/... [slashdot.org]
read your title before posting articles? (Score:3)
"risking drought risk"?
s/risking/rising
I can forgive grammatical errors in post bodies much easier than in the TITLE. It's one sentence, c'mon...
Once again (Score:5, Insightful)
A lower population would negate most of this issue. Less people = less water usage, not only for personal use, but for food production. We wouldn't need to aerate millions of acres of desert for crops [businessinsider.com].
Re: (Score:2)
A lower population would negate most of this issue. Less people = less water usage, not only for personal use, but for food production. We wouldn't need to aerate millions of acres of desert for crops.
You know we throw 40% of our food away, and also a significant portion of the the cropland is for fuel [civileats.com], right? We already don't need to do that.
Re: (Score:1)
Also, we export a lot of that food to places that just plain can't grow enough. Irrigating land to produce more food is how we're feeding 8 billion people. I guess we could drop nukes in WWIII until the population stabilizes but that seems like a bad idea.
Re: (Score:2)
Re: (Score:2)
Another leftist who wants to decimate the population.
If we weren't in this situation, we could conceivably support even more people than this, and with a higher quality of life. Sadly, nobody with money has been listening to calls for sustainability basically ever.
Re: (Score:2)
Most crop water is wasted.
Even stuff like Alf Alfa could be farmed with only 10% of the water it uses atm, if the warmers would care (some do) - but water is to cheap to care.
"Risking the Risk" ? (Score:2)