Slashdot Mirror
Internet Uses 9.4% of Electricity In the US
ribuck writes "Equipment powering the internet accounts for 9.4% of electricity demand in the U.S., and 5.3% of global demand, according to research by David Sarokin at online pay-for-answers service Uclue. Worldwide, that's 868 billion kilowatt-hours per year. The total includes the energy used by desktop computers and monitors (which makes up two-thirds of the total), plus other energy sinks including modems, routers, data processing equipment and cooling equipment."
Remember the article that more are browsing the web *instead* of watching TV? That would mean that TV power is going to PC's instead. (Except maybe for those who leave both on, and some PC's + monitor take more power than a TV)
Table-ized A.I.
So then I guess you are saying that since bittorrent comsumes about 50% of the internet bandwith it consumes perhaps half 4% of the power. Of course since bit torrent can be an edge network this might be more or less than 50% of power depending on if the edge is more or less efficient thant the backbone. My guess is that it is less efficient but that's arguable. One factor is if you want your home heated or not. That waste heat from the edge servers is heating homes and thus is an equivalent savings on the energy needed to heat homes. The opposite is true if you had the AC on. On the backbone all waste heat is working against the AC.
By the same token spam is also a major consume of world power. Now that would be a good reason to go against that!
If we assume most traffic is one the backbone and that the backbone scales as the number of servers running it. Then we only have a few more years before the power consumed by the internet will be larger than todays total power budget. This seems impossible. Ergo the traffic must be out on the edges. And there the scaling may be different with power.
Some drink at the fountain of knowledge. Others just gargle.
Furthermore, a large fraction of the remaining 1/3rd of power is servers. Many of them would be run even without the internet, most probably as internal servers for 1-800 phone reps.
The actual power attributable to the Internet is probably quite small. And certainly less than the gasoline and other motor fuels used in personal shopping/research/entertainment trips reduced by the Ineternet.
Please look at the total picture. Not some sensational part.
It would also be interesting to know how much energy the Internet saves. For example instead of people flying around they talk on VoIP or have a teleconference. Documents are emailed rather than having to be flown around the world. Music and movies are downloaded rather than people driving to the shops for a disk. Or is the Internet is promoting long distance relationships that otherwise just would not be?
The numbers do suggest that electronic equipment needs to be more efficient.
Or according to Einstein (and Google):
868 billion kilowatt hours = 3.1248 × 1018 joules
(3.1248 × (10 ** 18) joules) / (c ** 2) = 34.768089 kilograms
So keeping the current Internet running requires turning nearly 35 kilograms of mass into electricity.
Please direct all bug reports to
9.4% is probably way off, but here are some conversions/comparisons anyway:
868 billion kilowatt-hours per year = 10^11W=100GW
Space shuttle liftoff: 100GW
don't cut it off www.mgmbill.org
Trouble is that leaving computers running is arguably a rational business tradeoff. If a desktop computer draws 250 watts (and most don't average that high), and is left on during all non-business hours (assume one works only an 8-hour day and no weekends) that is 128 hours or 32 kWh or, at $0.10/kWh, $3.20.
If your entire employee cost (pay, bonus, worker's comp, medical, office-space, etc.) is only $60,000/year, an employee needs to save less than 10-minutes/week to break even.
One coder measured his own pretty high-end machine (including support for 3 monitors) at less than 140 watts when not doing heavy processing. This doesn't include the monitor which in most systems sleeps after a short period anyway. If we use 150 watts, a 9 hour day, and $100,000 employee cost then break-even happens by the time you have saved 2 minutes 15 seconds per week or less than 30 seconds per day.
Now if it takes 2 watts cooling per watt of usage then the benefits of shutting down are greater. But on the other hand, none of the office buildings where I've worked have metered power or cooling (except for custom auxiliary units) so from the tenant perspective, leaving the machine running has no impact on power or cooling costs.
Sure, for many, waiting for a computer to boot is part of the morning routine and provides an excuse to go fill the coffee cup. But if buildings metered power and cooling usage and if computers were made to save-state and swich off and back on like a light - or at least in just 1-2 seconds - people would be much more willing to power down not only at night but at lunch and whenever they aren't using the machine.
~~~~~~~
"You are not remembered for doing what is expected of you." - Atul Chitnis
"lowering the amount of power needed for the average home PC to operate."
And this will continue to change. People are becoming aware of resource scarcity, and want to insure themselves from rising prices. Witness the rise of cheap power meters such as the Kill-A-Watt. These took years to move over to 240V simply because they couldn't keep up with the demand for 110V items.
Something like a WRAP uses 5 Watts. Use it as a firewall/router/ADSL modem/traffic shaper, and it's going to be a cheaper and smaller solution than the typical 20+ Watt modem/router box.
Even CRTs have dropped in power usage compared to what they used to.
We are rapidly approaching the day when our computers will be fast enough for most tasks, the hard drive will be solid state, the system will be passively cooled and made from reliable parts that will last for decades, drawing minimal power. Any media that won't fit on the solid state hard drive can be stored on the spinning kind and plugged in as needed via USB/eSATA/firewire.
Intel probably doesn't want us to have these systems. AMD may or may not. Via certainly does, and you can bet that for pretty much everyone in the first world there is a market for several of these type of systems at a $300 price point or so. That may be a reduction in profitability for Intel, but it will be a massive new market for others, and getting easier to enter all the time.
If I have seen further it is by stealing the Intellectual Property of giants.
TFA states that 25% of the power consumed by computers goes toward powering local networking hardware, which is factored at about 20% of the total consumption of the Internet.
This means that a typical small office with 20 computers has local networking hardware consuming the equivalent of 5 PCs.
Sources cited in TFA state that each PC uses an average of (588kW/365.25/24*1000) = 67 Watts, which seems reasonable enough. But that (67*5) means that 335 Watts worth of network infrastructure gear are present in a 20-PC office, which is absurd.
I know that Cisco has been known to make some hot switches, but for fuck's sake. At my place of employ (a not-atypical 20-ish PC small office), we've just got a passively-cooled 24-port 3com switch which doesn't even get warm to the touch, two Linksys WRT54GL routers, and a cable modem.
High estimates for this scenario might be 15 Watts for the switch and 12 Watts for each of the other devices, for a total of 51 Watts for the entire network, or about 4% of that which is used by the PCs.
By these estimates, my own home network has a slightly worse ratio, at about 6%.
But even if we figure that everyone else in the world has a vastly more complicated routing and switching fabric than I portray here, I simply cannot envision this figure being beyond about 8% on average -- a far lower figure than the author's stated 25%.
This means that the total consumption of the Internet in the United States, as corrected, stands at about 8%, down from 9.4%. (Not much different expressed that way, until you realize that 1.4% of the total US electricity consumption really is a huge figure.)
If anyone else has any additional corrections to make, please do so. Your contribution helps keep the teh Intarwebs green.
Kid-proof tablet..
It's cool that somebody put a lot of money to investigate how much money we are spending in energy. Sarcasm aside, it is kind of cool to find out how much we really are using something. Well, at least through an estimate. Other than that, how is this news worthy. It would help out students to develop an argument for energy use or for a independent study to reference for their own report.
Should I be worried that downloading porn is becoming too expensive. Should we start reducing the amount of time on the computer so I can save energy. The weird thing about this is that it only accounts for 10% of energy use. What about the other 90%? According to the reverenced CIA Factbook it says that 71% of the energy produced comes from fossil fuels. How much of that is used in vehicles? I could understand coal and natural gas plants but how much of that is used for fueling planes, trains or automobiles? Secondly, businesses need computers to operate but they also need faxes, printers, not to mention light. I mean do we really have to have street lights on all night long. I mean there are millions of them and they are on for at least 8 hours each day. Again with the sarcasm, I apologize for that but come on.
It is unquantifiable because it is like piracy. Most of the software/media pirates would *not* have otherwise purchased the 'stolen' goods. Similarly, the present quantity of information simply would not flow if it was up to the USPS. You see this everywhere: Efficiency drives demand. I had an energy economics book (that I unfortunately loaned out lost record of forever) that made a compelling argument that essentially declared that the road to environmental/energy salvation was burning fossil fuels as fast and furious as possible.