The Footprint of Gmail: How Much Energy Would Deleting Email Save?
by Sami Grover, Carrboro, NC, USA on 02.17.08
Like so many others, this TreeHugger made the switch to Gmail a while back, and has never looked back. He immediately started making use of its massive storage capabilities, and stopped deleting anything but the most obviously pointless emails. His email inbox is rapidly approaching the 11,000 messages mark. Being the technologically-challenged individual that he is, it took him some time to realise that there might be consequences to these actions – namely data retained means storage space used, and storage space used means energy consumed. The more tech savvy reader is probably shaking her head right now, amazed at how it can take this long to realise such things, but the fact remains that many of us still see digital files as existing in some kind of limbo – if they are not using up paper, or taking up storage space in our filing cabinets, what kind of impact can they have? As Lloyd pointed out in his post on the paperless home, moving from paper to digital storage is leading to major increases in energy consumption, and Google itself acknowledges it is "as much of an energy glutton as heavy industry". But just how much energy are we using up by not keeping our Gmail inboxes in order?
This question lead to some interesting discussion among TreeHugger staff, with our resident computer-guru Mark Ontkush informing us that it is almost impossible to get accurate numbers. Apparently it used to be the accepted wisdom that 2MB of data required about a lb of coal to store annually, but this figure was based on a study from 1999 which has supposedly since disappeared from the internet (no mean feat, as Mark points out!). Whether this study was accurate at the time or not, the numbers stuck around in popular discourse. Apparently this was recalculated to somewhere between 10 and 20MB per lb of coal in 2003, but even those figures are likely to be way out of date by now. Whatever the exact energy use per unit of storage is, we do know that the internet uses somewhere around 3% of US electricity consumption – a pretty sizeable figure.
Mike also waded into the debate, pointing out that there are so many variables – from how much data is stored for how long and where, to how often it is accessed, from where, on which machines. Add to that the question of what the grid mix is for the locations in question, and the fact that from one request to the next, things can be routed to different places or gotten from RAM or HDD, at different data centers. Ultimately is seems that the only thing we can say in the end is this - less data equals less power by some amount.
Mike concluded by suggesting that the real question is opportunity cost: how much effort would we put into reducing emails, and would that effort be better used elsewhere? We’d love to hear our readers thoughts on this – meanwhile this author will be at least trying to get his inbox down under the 10,000 messages mark, though he’s going to try not to delete the discussion thread that started all this.
If you are using Gmail, don't forget you can add TreeHugger web clips to your account!
::Gmail:: via (frequent) site visits::
How much effort would it take? Basically none. It's the difference between choosing between Archive and Delete. Neither button is harder to press than the other!
More to the point. Floating existing memory bits does not add power consumption. Writing more memory or deleting it will cost energy. The loss occurs at sending and receiving and deleting not the storage itself. The storage is always powered regardless if it has available memory or not with no marginal increase with added data. Only accessing/writing/erasing data increases energy demand.
I think the amount of energy saved by deleting messages is negligible. For example, you've a notebook with 120Gb of storage You use only 20Gb. Does it mean that your notebook expense less energy than if it's stored with 100Gb of data? I don't think so... The energy used in operating the notebook should be the same regardless of the harddisk storing 20Gb or 100Gb of data.
Another example, your MP3 player can store 1,000 songs but you store only 400 songs. It doesn't mean that your MP3 player is more energy efficient with 400 songs compared to 1,000 songs.
I think the point is not on the messages stored but about optimizing the device resorces. Storing 100Gb of data on one large drive is more economical than splitting the data into two or more smaller capacity drives as it means having double the energy consumed by a single large capacity drive.
This is why server virtualization is so hot now as it allows companies to cut down on physical number of severs required and hence saving on resources.
Google should be the one to go green. A few megabytes of data won't make odds nor ends. If you don't fill it with emails, it'll be filled with search indices or advertising stuff.
There are a lot of things Google could do to go green. Buying a bucketload of green energy sounds like a nice greenwashing attempt to me.
I suspect that the act of going and deleting the mail (both client and server side energy expenditure there) would take more energy than letting existing email stay archived.
Hard drive efficiency and platter density are going up so fast that it doesn't matter at all in comparison.
Despite the variables, you've pushed me to organize and clean out my inbox, Sami... thanks!
BenSchiendelman, that might be the case for a single computer (where an extra 100mb doesn't mean you buy a bigger harddrive), but ultimately for such a big farm, that would mean an extra spinning disk, which I think was the author's point
Hitting the 'delete' button will not cause a hard drive to suddenly use less energy. It will continue spinning at a constant velocity with or without the data of that email on it's platters.
Think about it this way: your decision to delete an email (which, on average, is only about 2 kilobytes of data without attachments) may save that much space on a hard drive somewhere up 'in the cloud'. However, someone else's data will be there to take it's place.
The basic fact is that since it's inception, the demand for storage capacity and overall infrastructure of the internet has been increasing over the past 20 years or so. Instead of fretting over whether that extra 2 kilobytes of data will cause another piece of coal to be burned just so that hard drive will keep spinning, concentrate on the fact that research into the use of renewable energy sources to power data centers is already occurring (see Google.org) and there are 'green' web hosts out there (http://blog.dreamhost.com/2007/04/20/were-green/) ready for your business.
doesn't Google offset a lot of their footprint?
And aren't they also huge investors in Alt. energy?
So, by using GMail, aren't you actually contributing to Google's trust fund for renewable energy in some ways...
I mean, the tradeoff isn't perfect, and I'm sure the advertising revenue they get from my google account isn't going 100% into green purposes, but they do claim that they strive "[not to] be evil."
Are emails really deleted though? I'm not sure they are, I think that Google (and other ISPs for that matter) are obligated by law to keep them for some time... If that's the case then maybe the key would be not to receive big email messages to begin with.
The environmental marginal cost of storing additional emails and sending electronic communications decreases once the equipment is in place and as devices become more efficient.
EWout: I wouldn't be convinced that google actually deletes anything when you hit the delete key. Every piece of information they keep about you is money in their pocket (via ads), so they will keep all of it.
And when you buy space in 500TB chunks (or something like that), who really cares?
(As an aside, I have done work in a datacenter, and the best way to make them more power-efficient is to move them to the northernmost climates, because the AC bill is one of the biggest parts of a bill for them).
If the author really thought that deleting the messages from your inbox/gmail account was going to save energy shouldn't they consider how much energy it takes to actually run the machine that is accessing said gmail account? Couldn't you completely cut down on the energy output of a gmail account by just refusing to use a computer all together?
hey I wonder if you and your readers would be interested in this - if so, a link would work wonders..
http://peacharse.blogspot.com/2008/02/youre-not-only-one_10.html
True, but imagine if all documents now delivered electronically still needed to be transmitted via snail mail!!
Google is actually pretty agressive about energy efficiency. They build their own servers and power supplies so they can squeze every last bit of work they can out of a kilowatt-hour.
Storing that much email surely has a carbon footprint associated with it, but the footprint is probably smaller than if you had chosen to store it somewhere else, or on your own computer. Google has an army of techs who do nothing but wriggle around the data centers looking for ways to cut the electricity bill.
Yeah, they use a lot of power, but they also do a lot work. It would be illuminating, if you will forgive the term, to think in terms of watts per email, or watts per GB, or joules per query. On those terms, I would guess that Google actually does pretty well.
I'm a Live Hotmail user here, and I don't bother sending e-mail messages, unless I need to be formal or just drop a message to someone who doens't IM.
Otherwise than that, I just IM. I only use e-mail to get a notification from a site these days, although the switch to Windows Live Alerts might change that.
A hard drive draws somewhere in the neighborhood of 10 watts. (Less, actually, but 10 is a nice, round number). Google almost certainly oversubscribes its storage (i.e. they promise everyone several gigabytes, but they don't have enough storage for everyone to actually use that much) But, even if they didn't, a modern hard drive can hold more than 100x what they promise users, so your share of the cost is about 1/10th of a watt assuming no over subscription. Google probably pays almost as much in cooling costs as it does in storage costs and they almost certainly use RAID or some other form of redundant storage, which will increase power consumption because of extra drives. And there is the embodied energy represented in the drive itself. But the final cost will be a fraction of a watt.
We'd be better off putting our own electronics on power strips to cut off phantom power draws, only turning on lights when we need them, and buying less junk in general. Minimizing the amount of digital cruft that we keep around isn't necessarily a bad thing, but the above back of the envelope calculation suggests that it's not the first place we should start optimizing while there is plenty of low hanging fruit that can offer bigger payoffs.
(Although it's good to ask these sorts of questions -- the answer isn't always "Too little power to really matter")
New energy efficient drives use something like 4 watts when idle. If you leave it on all day that is 4*24/1000 = .096 kilowatt hours per day. Of course you could shut your drive off most of the day. That may get you down to .033 kilowatt hours per day. Now when you consider that email will take up about one percent of that disk you see that email will cost about .000033 kilowatt hours per day. That could be provided with a solar cell the size of a postage stamp. I think there are bigger fish in the sea. Best to buy one big energy efficient drive for you home. Use it as a back up. Do a back up and shut the thing off.
Great article, but this commenter hates it when bloggers write in third person.
What the hell do you need 11,000 e-mails for?
An e-pack rat?
You guys have got to be kidding.
Email storage (as in the text part) is an infinitesimal drop in the data storage ocean.
How many terabytes of of images, particularly video are there out there do you suppose?.
We have security videos, traffic cams, YouTube videos, satellite images, deep space images (from the Hubble), piles upon piles of family and vacation photos, and just endless porn. Now all of the movie catalogs of all the studios and every TV show ever filmed will be online. They're out there now bouncing around in bit torrents already. More data is streaming in real time every second.
Then there is the data storage of industry and the governments: tax records, legal documents, all sorts of databases for personnel and sales and parts and health records and bank transactions not to mention spy stuff like the NSA recording everything all the time.
How many VISA transaction are they a day do you suppose? How big of an audit trail does each of those transactions leave across multiple computers? How long are those records stored?
How about airline reservations? Some of the biggest commercial databases in the world do nothing but manage airline reservations.
The most insane thing I seen lately is telephone records. I have four teenagers. Every call, every text message, tracked, recorded, listed on the bill (I used to get a 30 page bill every month until I went paperless). Those records are kept for a long time. They can be subpoenaed. How many cell phones out there? How many calls or msgs per day?
At this point you would have to disengage from 'modern' life completely to avoid creating all these database records. Don't get a drivers license, don't fly, don't have a bank account, don't pay taxes, no phone calls. If you run away into the wilderness it will create a database record when your friends report you missing, and It would still be hard to avoid being photographed by a satellite or Google Street Level.
I guess you need to balance all this electronic storage against the cost of printing it all on paper or plastic film and maintaining it in a climate controlled environment. The density of electronic data is much higher so presumably it can be stored more efficiently.
For a great future vision of all this (or nightmare depending on your POV) read Vernor Vinge's Rainbows End.
I think the true point here is that the more emails stored, the more storage capacity Google will have to add to its data banks. This is not a continuous progression, rather it will occur in "steps" or "bursts." When a threshold of users is reached, Google will purchase more capacity. When this capacity comes online, that is when the energy increase will occur, not when you delete the message.
The idea is the same with almost every "cause" related conservation method typical consumers choose to support. Want to reduce methane emissions and water usage by not buying steak? You may be surprised that your individual choice has absolutely ZERO immediate impact because that tantalizing slice of ruminant rump you painfully pass up has already had its impact. What you are doing, however, is influencing consumer demand. If say 10 - 15 people make this same choice consistently for a sustained period of time, maybe 6 - 12 months later, the price of steak will fall by 1/300,000 of a percent. This will cause one less steer to be purchased for breeding a few months later. This will in-turn reduce the number of meat cow births to fall by 1 - 2 head that year. Then 3 - 4 years later, there will be that much fewer cows in circulation. So about 5 years after you make your choice to stop eating steak, the ripples it creates in the open market will have some type of effect. Only then will you save that 2,000 gallons of water you were hoping to conserve right there in the supermarket.
The point of the above paragraph is that there is so much market momentum pushing back against individual user choices that it requires a large number of people to make a difference and that even then there will be a multi-month to multi-year lag before it has any sensible impact. And then you have to understand that this impact is going to cost jobs, which when weighed against the environmental impact may be acceptable.
So, don't think individual buying choices are going to make an immediate impact. By all means, do them. But expect that markets must reach equilibrium before the true impact will occur.
As for the Google data farms, delete your emails or download them to your own computer via. a local email client such as Thunderbird of Outlook. If 100,000 other users do this too, then Google won't have to buy that extra 100 TB sector 6 months from now and you will have been part of a movement to stop Google from buying energy on your behalf.
Lucas
In addition to the minuscule watt per byte of operating a hard disk, as has already been mentioned, let's go over the cost of accessing 11,000 emails. Even if you only had 1 or 2 email that you read day in, day out, over and over again, it would cost virtually exactly the same as loading and searching through 11,000. A search of your email does not actually operate on all the data of those 11,000 messages, so it's not like the disk has to chug along reading all those bits. A much smaller, highly efficient index is used. Using an index is actually disadvantageous if you've only got a few emails, and unless you're an algorithmic expert at Google, it's anyone's guess how they right-size those indexes - so you can't say at what threshold there will be no more discernible savings in computing effort to search through. But with an index, your emails can sit on a disk that's completely powered off for all that anyone cares, but you can still find it and retrieve it if you want to read it again. Google only needs to predict what data you are most likely to look at during your session and cache that for you - so some of your emails along with your index file can quickly be loaded up into a main server when you sign on and the drives on which your data actually reside can go back to sleep and power themselves off. Only the people currently signed into their gmail accounts, actively looking at emails, need to have powered-on hard disks.
If you are really hell-bent on saving minuscule amounts of energy, try turning off some of those advanced labeling features, which seem to require extra processing power - access your email the old fashioned way. What's even better? Sign out when you're done using it, which will prompt the server to free up some of the memory you're using on it's active drives and ram. Even better? Shut off your own computer.
How much energy are we really talking about here? I would venture to say that the energy used to produce a sheet of paper and fill it with text is much more than the cost of storing the same amount of information on a hard drive, even if that paper lasts 1,000 years and you have to replace the drives as they burn out. The amount of data on a single terabyte hard drive could potentially hold 500 million pages of typed text.
In addition to the minuscule watt per byte of operating a hard disk, as has already been mentioned, let's go over the cost of accessing 11,000 emails. Even if you only had 1 or 2 email that you read day in, day out, over and over again, it would cost virtually exactly the same as loading and searching through 11,000. A search of your email does not actually operate on all the data of those 11,000 messages, so it's not like the disk has to chug along reading all those bits. A much smaller, highly efficient index is used. Using an index is actually disadvantageous if you've only got a few emails, and unless you're an algorithmic expert at Google, it's anyone's guess how they right-size those indexes - so you can't say at what threshold there will be no more discernible savings in computing effort to search through. But with an index, your emails can sit on a disk that's completely powered off for all that anyone cares, but you can still find it and retrieve it if you want to read it again. Google only needs to predict what data you are most likely to look at during your session and cache that for you - so some of your emails along with your index file can quickly be loaded up into a main server when you sign on and the drives on which your data actually reside can go back to sleep and power themselves off. Only the people currently signed into their gmail accounts, actively looking at emails, need to have powered-on hard disks.
If you are really hell-bent on saving minuscule amounts of energy, try turning off some of those advanced labeling features, which seem to require extra processing power - access your email the old fashioned way. What's even better? Sign out when you're done using it, which will prompt the server to free up some of the memory you're using on it's active drives and ram. Even better? Shut off your own computer.
How much energy are we really talking about here? I would venture to say that the energy used to produce a sheet of paper and fill it with text is much more than the cost of storing the same amount of information on a hard drive, even if that paper lasts 1,000 years and you have to replace the drives as they burn out. The amount of data on a single terabyte hard drive could potentially hold 500 million pages of typed text.
The enery saved is too small to matter
Oh please. Aren't there much bigger REAL issues to worry about? The size of the average email account is going to be incredibly miniscule compared to the massive amount of data being stored by Google. Besides, from what I've read, Google does not use hard drive storage - everything is cached in RAM and replicated across redundant servers. I'll bet that the simple act of checking your email (causing web pages to be served up from Google's servers and rendered on your local browser, all the while travelling through your ISPs systems) expends more energy than the lifetime storage of the data sitting in your email account.
If you're really concerned about your energy footprint, how about checking email less often (and powering off your computer when not in use)? The less time you spend online, the more energy you'll save.
This talk about rationing your email to save bits is just ridiculous on so many levels. At first I thought I was reading a spoof piece a la The Onion, then I realized the author was actually serious...
so if this is the equivalent of high hanging fruit, where is the low hanging stuff? Oversize flickr acct.? The mountains of spam that servers process? Using wifi when ur sitting next to a physical port? Second life, WoW, and increasingly power hungry videogames?
But let's not forget to compare each of these to their IRL counterparts. Snail mail, printers/fax, filing cabinets, and travel for in person games of DnD!
this presumes that google has dedicated servers and hard disks allocated to gmail. which is NOT the case. so deleting your mail really doesn't help anything, consumption-wise.
For what it's worth: Empty the spam folder regularly. For your inbox, set up a "junk" label and apply it to anything that is, in your opinion, "junk." Every now and then, call up all messages labeled "junk" and delete them. With a bit more effort you can set up a filter to add the "junk" label to certain recurring emails automatically.
My $.02 here - take it or leave it!
Spare Google from the burden and download the mails to your (or a) computer, which runs anayway if you need access to your email. And you won't lose emails by faults. And they happen. All they will say is sorry. But your 11000 emails are gone.
But how much energy gets used every time you post to a blog? (Or add a pointless comment!) Every blog post you make is accessed, archived and cached across hundreds of thousands of computers.
Worry about the big stuff first. We look like crackpots (and rightly so) when we start fretting about the energy consumption of a cluttered email box.
One subtext to my earlier post is that humans have a fascinating pack rat psychology and an obsession with recording everything imaginable.
Another subtext is that if you look at the way people use cell phones, text messages, IMs, email, etc they really really want/need to make contact with other people all the time. How much of that messaging has essentially zero information content other than "I am here. How/where are you?" Just relationship maintenance signaling like whales seem to do.
These two aspects of human nature combined and created an explosion of personal data creation and storage as soon as the technology was available to consumers.
I think it reveals a basic human need being fulfilled amid what would otherwise appear to be the sterile techno/corporate landscape of modern urban life.
My third (and final) subtext is that all of this is still a pretty small piece of the overall background data storage/energy consumption framework created by the global data communication and storage network.
As for me, all of my professional business and a pretty big chunk of my personal life is carried out in and via these media. I like to save everything because it creates an accessible database of where I have been and what I have done and how others reacted to it.
Almost invariably if I delete something I soon regret it.
A very fascinating project with respect to this is Gordon Bell's Lifelogger project: http://www.semanticvoid.com/lifelogger/
wherein he attempts to document in fully annotated audio/video and interlinked documents every moment of his life.
I guess one caveat to this high horse position is called for here, I do use SPAM filters at various levels that auto-delete for me email and messages I don't think I need to see or save...
I read most of the comments and I agree with many that suggest deleting some of your email messages would not be equivalent to reducing the energy used (in storing them). The argument there revolves more around how much are we storing/deleting and in what ways.
However I think the original intent of this article was to look at this from the macro level; not one person on their sony vaio saving all their emails in Gmail...but a population of Gmail users, clogging up servers with useless communication that they can/should live without. If they should live without it, why? Because maybe it reduces the space needed (for hundreds of thousands of users) by a roomful of servers? Etc.
Can't we just go after the schmucks who write spam and delete them?
Eventually we will switch to using FLASH NAND, which doesn't require as much nrg, though I would think that data storage does require NRG, that stuff has to be cooled, and the speed in which it's accessed, a larger amt. of memory would inevitably use more NRG. even if the equation is now 20 gb = lb. coal, thats still quite a substantial amt. but that would mean that I myself am using less then a lb. of coalper year to run my internet lifestyle.
Eventually we will switch to using FLASH NAND, which doesn't require as much nrg, though I would think that data storage does require NRG, that stuff has to be cooled, and the speed in which it's accessed, a larger amt. of memory would inevitably use more NRG. even if the equation is now 20 gb = lb. coal, thats still quite a substantial amt. but that would mean that I myself am using less then a lb. of coalper year to run my internet lifestyle.
Would Treehugger produce an honest article on external harddrives which use the least amount of energy while in use and idle?
Thanks!
Yahoo! has green data centers and reduces his footprint via green electricity projects around the world, and his mail data storage is unlimited!!.
Yahoo! rules!!!
look:
http://ycorpblog.com/2007/10/21/making-good-on-our-promise/
http://ycorpblog.com/2007/04/17/dont-even-leave-a-footprint/
http://green.yahoo.com
I had noticed a comment about Google being dedicated to alternative energy, and you can read more about it at google.org. Their RE
As for this business, I think the idea of cutting down on any sort of useless existence is a good thing. I mean, sure it doesn't make much of an impact, but to say "if there isn't email filling the servers, then something else will" kinda misses the mark. if the other stuff is going to exist no matter what, more email will just require more server space, and in turn, more energy consumption. no?
deleting emails to save energy is like planting grass as a CO2 mitigation scheme. Not worth the time.
This blog entry made me realize how one company can assess a growing trend; transition from paper to digital storage,