Re: Cost Justifications

[Evan Harris <auctionitis@xxxxxxxxx>]

Thanks for such a cool (pun intended) answer to what was a just a
little bit of a tongue in cheek question.

Reading your answer, I have another question:

The figure you quote of 3413 BTU per input KW - is this dependent on
anything else ?

What I am really wondering is: is there any difference between intel
vs power architectures - does one generate more heat than another ? Or
are the differences in heat really about the multitude of boxes that
tend to accumulate round Windows/intel applications ?

If I ran an 8 way dell server with all my windows applications
virtualised would it run hotter or colder than an 8 way power box all
other things being equal ? Assume for instance that they are both
using the same IBM SAN for storage.

Thanks for taking the time with the previous answer .

On Thu, Mar 10, 2011 at 3:03 AM, DrFranken <midrange@xxxxxxxxxxxx> wrote:
> OK I wrote a certification question on the ratio between input power and
> output heat and got booed off the stage. SO I will simply tell you the
> answer. For each KW of input power (1000 Watts) you will get 3413 BTU of
> heat output. Now since a teeny tiny little bit of that 1000 watts went
> into creating LED light (a VERY teeny bit) and some into noise generated
> by the fans you might get a measly 3410 BTU of heat.  Point is that it's
> a straight line relationship between input power and output heat.  It's
> fairly common for HVAC guys who know there stuff to ask what the total
> power input to the data center is, what the percentage load on the UPS
> is, and use that number as a basis for calculating cooling requirements.
> If you put 100KVA IN you will need to get 341,300 BTU out somehow.
>
> Now how much electricity it takes to remove those BTUs can vary A LOT.
> In very cold climates, outdoor air can be drawn through filters to cool
> the place (devices called 'economizers') and that's cheap. In hot
> climates very large amounts of A/C are used to cool such rooms and that
> cost can rival the power of the systems being cooled.  In the FrankenLab
> from about October through March 100% of the heat generated is turned
> into domestic hot water and then heat for the house.  In the summer it
> ends up in the swimming pool or down a dry well. Very economical indeed.
>
>     - Larry 'DrFranken' Bolhuis