The unix hackers at the Technion did that to the Vaxen 11/7XXes;   didn't
trust the walk-in UPS, because the mainframe and Vax 9000 would drain it to
the dregs. Couldn't do the same thing for the Sun 4/XXX because no source
license.

Up hill, both ways, in the sand.
Simon

On Mon, May 10, 2010 at 4:47 PM, Casey Bisson <[log in to unmask]> wrote:

> Or be like Google, give up on UPSs, and just attached a battery to the DC
> side of each server's power supply.
>
>
> http://arstechnica.com/hardware/news/2009/04/the-beast-unveiled-inside-a-google-server.ars
>
>
> On May 10, 2010, at 4:41 PM, Simon Spero wrote:
>
> > At least it wasn't  a "totally transparent"  UPS test scheduled for the
> > Thursday of Thanksgiving weekend.  My personal philosophy is that every
> rack
> > should have its own UPS separate from the data center one, with enough
> > capacity to keep going through blips,and handle a clean shutdown if
> > necessary.  That way,  when the ops team messes up, far fewer sysadmins
> get
> > their weekend ruined.
> >
> > Of course, the real problem is that too many people are writing
> unoptimized
> > code in energy-inefficient languages like ruby and PHP, which require far
> > more servers, and far more cooling, to do the same work as properly
> written
> > code.  If carbon emissions should turn out to be a  strong forcer  of
> global
> > warming, then we can clearly say that every time you write  PHP, Phil
> Jones
> > kills a polar bear. Please, think of the polar bears.
> >
> > Simon
>