UPS (Uninterruptable Power Supplies)... kinda long.

root at MCIRPS2.MED.NYU.EDU root at MCIRPS2.MED.NYU.EDU
Mon Jul 23 03:01:17 AEST 1990


I have had people at the med ctr equip XT class PC's with UPS that cost an
order of magnatude more than the machine that they protected.  My feeling is
that they really needed a more reliable machine, a more powerful machine, and
should stop using pc's and move up to some type of reliable unix system.  Our
ncr unix system has a reliable powerfail deamon, and a power supply with large
caps and a motorcycle battery.  The deamon monitors the powerline at the inlet
plug, and can signal the machine to start saving its state when power gets
low.  This means sync'ing the file system, saving/flushing the terminal io,
and when the system is sunc(synked), spin down the disks.

Many workstation class computers have NOT had good power fail properties, and
this is a shame.  You do NOT need expensive ups if your machine can recover
from a power interruption.  I think that a better power deamon is required in
engineering workstations (sgi's) when you run programs that take more than a
day.

I have been in computer room design projects where the designers think nothing
of dropping 100,000 dollars on ups and special power, when modern machines
should NOT need it.  I would rather have the money for machine, rather than AC
power.

UPS would be more of a critical issue in the Power Server series disk farms,
but a PI should also tolerate having its power cord yanked from the socket.

We have terrible power at the medical center, and I have never had the need to
spend money on ups's for our machines.  What is more silly is that I can get
lots of money from the electrical department for UPS systems that I don't
need, and I am choking on lack of money for more powerful computers.

All that I would want from a iris workstation is that it not loose files, and
unclosed file buffers when it looses power.  I would also like to optionally
save a running jobs memory state so that I could resume long programs after an
interruption.

I don't feel the need to use my iris in a power failure by candle light.

--
+-----------------------------------------------------------------------------+
| karron at nyu.edu                          Dan Karron                          |
| . . . . . . . . . . . . . .             New York University Medical Center  |
| 560 First Avenue           \ \    Pager <1> (212) 397 9330                  |
| New York, New York 10016    \**\        <2> 10896   <3> <your-number-here>  |
| (212) 340 5210               \**\__________________________________________ |
+-----------------------------------------------------------------------------+



More information about the Comp.sys.sgi mailing list