Slashdot Mirror
Energy Use From Wireless Networks Will Dwarf Data Center Use By 2015
angry tapir writes "New research (PDF) from an Australian university argues that increased carbon emissions from powering data centers aren't the biggest environmental threat from the growth of cloud computing. Instead, the problem is the Wi-Fi and cellular networks increasingly used to access cloud services. By 2015, the energy used to run data centers will be a 'drop in the ocean' compared to the energy used to power wireless access to services. By 2015 the energy consumption associated with 'wireless cloud' will reach 43 terawatt-hours, compared to 9.2 terawatt-hours in 2012 (an increase in carbon footprint from 6 megatons of CO_2 in 2012, up to 30 megatons of CO_2 in 2015). Data centers will comprise only 9 per cent on this increased energy consumption, compared to up to 90 per cent for wireless access."
The amazing article doesn't say exactly what the "wireless cloud" is. Wifi, perhaps, or do they just mean ramping up mobile phone networks adding capacity and transmitters?
Naturally they didn't bother to compare the power consumption of wireless to having multiple wired connections to everything. Wireless devices themselves tend to use less power than desktop PCs so probably more than offset the energy required for transmission of data.
It also uses the often heard but stupid reasoning that A is bad, but B is far worse so A isn't so bad any more.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
...as they will be used more and more to provide a near-instant (Twitter-like) service that turns PDFs to HTML as soon as one is linked somewhere on the web
If computers were people, I'd be a misanthrope.
The article has very little detail in it.
It really sounded like they took the curve of data center power reduction and forecast it out, then took the amount of energy used by every corporate and home wireless device in the country(world?) and every cellular network (and it's devices) and projected them out.
Apples and oranges.
It also sounds like they felt a dire need to stick "Cloud" into the article for no other point than to raise the headline value. The article did nothing at all to convince me that their predictions really relate to cloud computing any more than anything else.
Warning: Teh poster of this messaeg is lysdexic
(an increase in carbon footprint from 6 megatons of CO_2 in 2012, up to 30 megatons of CO_2 in 2015).
World CO2 emissions for 2011 was estimated to be over 33 thousand million tons. I will not be losing any sleep over this tiny bit of manufactured melodrama.
http://en.wikipedia.org/wiki/List_of_countries_by_carbon_dioxide_emissions#List_of_countries_by_2011_emissions_estimates
Is a drop 9% of the ocean.. Stupid "hyperbolists"
A factor of 5 does not justify the use of the metaphor "drop in the ocean". I wish people would reserve the use of metaphors for where they belong. "Metaphor inflation" just makes it harder to express yourself when using metaphors appropriately because nobody can trust that you mean what you say. The metaphor I would use for a factor of 5 is "dwarfs" or "pales in comparison". Drop in the ocean should be used when one is infinitesimally insignificant next to the other, which is not the case here.
Are they talking about power going to WiFi access points and routers and cell networks, or are they throwing in powering every single cell phone, tablett, and laptop used to access these networks as well. The article just doesn't specify, but I highly doubt that the energy to run a couple of dozen WiFi access points (or hundreds if you are a REALLLY large business) is ever going to be anywhere near what it costs to power servers, disc arrays, tape backup systems, routers, switches, and PBX systems.
With this being from Austrailia, for all we know, they are refering to covering the outback with cell towers and how much power that will take versus the power needed to locate a data center out there. The article is very vague.
The executive summery in the PDF pretty much states the same.
If you start digging down into the article PDF, I did find this:
This white paper presents a detailed model that
estimates the energy consumption of cloud services
delivered via wireless access networks in 2015 taking
into account the broad range of components required to
support those services, including data centres and the
telecommunications networks. The model is based on the
expected up-take of wireless cloud services and forecasts
of the telecommunications technologies that will underpin
wireless cloud services in 2015. This estimate uses an
incremental energy calculation that is based on a scenario
where wireless cloud traffi c is part of many other traffi c
fl ows through the network and data centres. Wireless cloud
traffi c is carried through a network that is already carrying
a large amount of traffi c, with wireless cloud traffi c being
about 20% of mobile traffi c and approximately 35% of data
centre traffi c [2,4].
So this makes more sense, but is seemingly talking in circles. The power required to power cell towers, wireless networks, and the datacenters to support them is going to be greater than the power needed to support data centers. Um, thanks.
What is worse is that the white paper reads like a lazy college student's attempt to present facts without really understanding the facts. I used to throw papers like that together in college. You have facts that you know you need to present in your paper, but you have no clue what they really mean, and almost get to the point where you are copying and pasting tidbits into your paper (and just citate the hell out of it).
I like to measure things in HydroQuebec's. That is, measuring usage relative to the installed capacity of my province's power company.
Hydro Quebec installed capacity: 36,971 MW
Hydro Quebec max generation per year: 324 TWh/year
TFA's power usage: 43 TWh/year
So TFA says in 2015, GLOBAL wireless transmission energy usage will be about 13% of HydroQuebec's capacity... Suddenly doesn't seem so much.