The Power of the Internet

[Charlie Monoxide]

We've all heard this tired description of the internet. It is always used in the context of the empowering power of the information it provides.

My question is, what exactly is the power REQUIREMENT of the internet? We have the many backbone, satellites and server farms. Many sites are also running on desktops in peoples homes and businesses. How much power (in kilo/megawatts) does the internet consume in an average day? Could we save a certain amount of pollutants from the dirty (ie coal and gas) generators, if we just picked a day of the week and not logon?

Charlie (too addicted to the internets to quit) Monoxide

[tsg]

We've all heard this tired description of the internet. It is always used in the context of the empowering power of the information it provides.

My question is, what exactly is the power REQUIREMENT of the internet? We have the many backbone, satellites and server farms. Many sites are also running on desktops in peoples homes and businesses. How much power (in kilo/megawatts) does the internet consume in an average day? Could we save a certain amount of pollutants from the dirty (ie coal and gas) generators, if we just picked a day of the week and not logon?

Charlie (too addicted to the internets to quit) Monoxide

I remember a story (disclaimer: I make no claim to it's veracity) about George Bush suggesting that we could save energy by not sending so much email.

The number of users does not significantly affect the power requirements of any particular machine.

[Upchurch]

I was just reading about this in eWeek.

How Green is IT's Future? (http://www.eweek.com/article2/0,1895,2168457,00.asp)
The EPA estimated that the IT industry consumed about 61 billion kilowatt-hours in 2006—about 1.5 percent of the total electricity consumed in the United States—at a cost of about $4.5 billion. Power consumption in the industry could nearly double by 2011, the report said. Federal servers and data centers accounted for about 10 percent—or 6 billion kwh—at an electrical cost of about $450 million.

That probably doesn't account for everything, but it gives you a ballpark estimate.

[Here_to_learn]

The number of users does not significantly affect the power requirements of any particular machine.This article (http://blogs.zdnet.com/storage/?p=165)seems to indicate that the load of a server does affect the power consumption on a modern machine:


Modeling CPU scaling under data center loads, Google found that data centers could see savings of 15-25%, depending on how aggressively it was used. They also found that I/O bound servers benefited less than compute-intensive workloads.

and
If idle power were only 10% of peak power, an enterprise data center could save 50% on its power. Even Google’s well-behaved apps would see savings in the 30%+ range. We can expect to see Google push the non-peak power efficiency issue pretty hard, and since they spend some $500 million a year on servers, I expect vendors will listen.

Now, from my understanding most web servers would have high I/O loads rather than CPU loads, so the savings would be lower, but still there would be savings if less people would access them.

My guess would be that quite a lot of the power consumption that Upchurch quoted would be related to desktop/end user computers, and they do certainly consume less energy if not used, even taken into consideration that they do consume some energy in what most people think is "off".

[strathmeyer]

Didn't we have a "weight of the Internet" thread?

[tsg]

This article (http://blogs.zdnet.com/storage/?p=165)seems to indicate that the load of a server does affect the power consumption on a modern machine:
Modeling CPU scaling under data center loads, Google found that data centers could see savings of 15-25%, depending on how aggressively it was used. They also found that I/O bound servers benefited less than compute-intensive workloads.


The 15-20% quoted above comes from power saving techniques that aren't in the vast majority of computers in use today.

and
If idle power were only 10% of peak power, an enterprise data center could save 50% on its power. Even Google’s well-behaved apps would see savings in the 30%+ range. We can expect to see Google push the non-peak power efficiency issue pretty hard, and since they spend some $500 million a year on servers, I expect vendors will listen.

The sentence just prior to this paragraph says, "Google found that idle systems power never dropped below 50% of peak load, while ideally an idle system’s consumption would drop to zero." In other words, if idle power were only 10%, they could save that much, but it isn't, so they don't.

Now, from my understanding most web servers would have high I/O loads rather than CPU loads, so the savings would be lower, but still there would be savings if less people would access them.

My guess would be that quite a lot of the power consumption that Upchurch quoted would be related to desktop/end user computers, and they do certainly consume less energy if not used, even taken into consideration that they do consume some energy in what most people think is "off".

I didn't say there weren't some savings, just not enough to be significant. The average homeowner will be able to save a lot more energy simply by turning off lights when he doesn't need them.

[slyjoe]

Didn't we have a "weight of the Internet" thread?

Yes - but it seems that in the latest reorganization of the forum that thread has entered the bit bucket of no return.

These discussions, to me, seem to turn into a "how many grains of sand are on a beach" thing. We know it's finite, but calculating anything meaningful is kind of a waste of time - IMHO :)

[Here_to_learn]

The 15-20% quoted above comes from power saving techniques that aren't in the vast majority of computers in use today.Agreed! But I would think that the average server on the Internet would be replaced in 2-3 years, so we will probably see a change in the next couple of years.I didn't say there weren't some savings, just not enough to be significant. The average homeowner will be able to save a lot more energy simply by turning off lights when he doesn't need them.Yes, there are several way to save power for the average house owner. In many regions either the cooling down or the heating up of the house is the majority of the power consumption, and then better built houses, or better insulation on the current house can be a major saving.

A computer/TV/appliance on standby can consume energy on par with a lamp or two, and turning them off instead of having them running all the time it can come up to quite a saving.

But as you say, just shutting down the lights, or sometimes something as simple as using a directed (but lower powered) lamp can make big savings over time.

[Charlie Monoxide]

My question has been answered by Slashdot: http://hardware.slashdot.org/hardware/07/09/27/2157230.shtml

Apparently Internet Uses 9.4% of Electricity In the US ....

Charlie (I love Slashdot) Monoxide

[strathmeyer]

Yes, but how much does it weight?

[Charlie Monoxide]

Yes, but how much does it weight?Be patient and the gods of info at Slashdot will answer your question eventually ....

Charlie (sleeping better knowing the power of the internets) Monoxide

[tsg]

My question has been answered by Slashdot: http://hardware.slashdot.org/hardware/07/09/27/2157230.shtml

Apparently Internet Uses 9.4% of Electricity In the US ....

Charlie (I love Slashdot) Monoxide

Does that include the energy used for the ensuing argument?