Your experience is completely counter to ours. Your utility power must
very stable, or you have experienced some UPS installations with bad
engineering designs.
We have evolved over approximately the last 15 years from a single UPS
system with no generator backup covering just the main computer room to
what today is a triple UPS system (any two of which are sufficient)
covering the main computer facility and all workstations in the
corporation, backed by stand-alone generation capacity sufficient to run
the entire corporate headquarters. This system now saves our tail from
downtime at least twice a month that I know of (just by observing light
blinks during normal working hours) and several winters ago kept us
mostly functional when we were without utility power for over a day
after a major ice storm.
I can recall the days before a UPS, whenever a major electrical storm
passed through it wasn't a question of if we were going down, only when.
When storms came at night, I often didn't bother waiting for the phone
to ring, just got up and got dressed and called to verify we were down
before leaving for work. All it took was a fraction-of-a-second power
outage, and it would take an hour to get all the hardware and software
fully functional again, provided no permanent hardware failures were
caused. Those days are thankfully long gone. The other thing we
noticed was isolating mainframe and peripherals from utility power
glitches significantly improved MTBF on the hardware.
We started with just UPS and no generator backup because 99% of utility
outages at that time were very short. Once users grew accustomed to
better availability, then eventually management saw the need to address
longer outages and that capability was added when we built the current
corporate headquarters building. Most of our outages are still short,
but we have at least several every year that are long enough for the
generator to cut in.
We have had cases where UPS failures have caused outages (including an
infamous case of a backing workman, before adequate butt-guard covers
were installed on a critical UPS switch), but we've never had a year
where they weren't a net benefit.
It is critical that UPS and generator design be properly integrated with
the building electrical system: That the UPS design support reliable
automatic and manual bypass capability that will remove it from the
picture before internal problems are seen downstream and to allow non
disruptive maintenance; that the UPS have reasonable self-diagnostic
capability and maintenance procedures so your first warning of failing
batteries isn't the UPS failure when needed; That generator controls and
switchover have ability to properly synchronize with utility power so
non disruptive switching can be done even if the UPS is non functional.
And perhaps most important, operating procedures must be clearly
understood and documented to avoid outages from dumb procedural
mistakes. With this configuration, regular test of generator capability
is possible, because critical systems are protected by UPS should there
be some problem with switchover; and at other times the generator can be
used to protect from utility glitches while maintenance is done on the UPSs.
Phil Payne wrote:
I worked for four vendors - ITEL, NAS, BASF/Comparex and Amdahl from 1978 to
1992. I have
never seen a situation in which the presence of a UPS actually improved matters
- they are a
much greater source of problems than the public electricity supply. My
definition (below) is
the honest result of years of experience.
And I HAVE actually seen a customer executive cry.
http://www.isham-research.co.uk/dd.html
--
Joel C. Ewing, Fort Smith, AR [EMAIL PROTECTED]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
