[clug] Re CO2 footprint of Searches: Storm in A Tea Cup or Deep Green Issue?

[jm]

If we're talking energy shouldn't we be talking Joules, ie, W = J/s, or 
rather J per unit of consumption?

The think to keep in mind here is that these thing have huge fixed 
energy costs and very low variable energy costs. To the example of the 
printer which consumes 24W in standby and 25W while printing. Lets 
assume that it takes 1 second to print a page then thats 25 J/page, but 
if you only print one page per day but leave the printer in standby for 
the full day:

second per day = 86400
Energy per day in standby = 86400*24 = 2073600 J
extra to print page = (25 - 24) = 1J

or 2.07MJ to print a page.

In other words if you don't need you printer turn it off. The same 
argument shows why public transport is a  poor use of energy when under 
utilised, but you can't turn off public transport on turn it back on 
when needed!

Returning to the energy cost per bit of data if these routers and 
switches are being well utilised, ie have a reasonably high average 
load, then it may not matter that the idle enegry consuption is high as 
the increase in energy consumption even at full load in negligible. To 
see if this were the case, you have to know the total energy consumed 
idle and at a typical work load plus what that typical work load was.

Strangely enough, by you would be better of using the existing hardware 
that adding in extra, but more enegry effiecent hardware. (Unless you 
decommissioned the old.)


Jeff.