Re: UPS hardware and software
On Wed, 21 Nov 2001 10:13:11 -0600, Dimitri Maziuk wrote:
>* Mark A. Bialik (mbialik@infinityhealthcare.com) spake thusly:
>> Liebert equipment is true line-interactive hardware... you are always
>> running from battery (which is continuously charging). Try running
>> APC's in an environment with a backup power source (Diesel/Natural Gas)
>> and see how they handle the switch-off... maybe you'll get lucky, maybe
>> not.
>
>We have APC SmartUPSen on the servers and various small APC UPSen
>on the workstations (mainly power-board style OfficeUPS). There's
>a generator on one of the circuits feeding the servers. So far
>the UPSen handled the switch-over just fine.
We're running about 70 APC UPSes. They are a mix: mostly
SU-1400RMs, some -1000s, -700s, -600s and even a few
BackUP 250s and Pro-280s.
The APCs were relatively inexpensive, had good software
support (under Windows), did more internal monitoring than
most UPSes at the time and could be bought just about
anywhere. Tech. support from APC has also been OK.
They support PCs running HVAC software in machine rooms.
If main power fails, the UPS only has to keep the equipment
alive until the BU generator spins up. We haven't had any
problems to date with running them from the generator power.
When they work, they seem to work well.
I'm am reconsidering them for our application though. They are
also line-interactive. This means that they are always "working"
the batteries and I think this tends to shorten the lifetime of the
pack. The SUs do an internal pack test roughly every two
weeks. If they have a shorted cell in a battery when the test
runs, they have a tendancy to dump the load, so you _could_
loose your server to a UPS test even with no other trouble
condition than a bad pack you haven't gotten around to replacing.
The monitoring for the SUs over a serial line is good (apart
from the weird cable pinout) but in some cases all I wanted was
a contact output. They don't provide one: only an open-collector
that you need to cook up a circuit for. This may not be a problem
for one or two, but try 70.
The 3U rackmounts have user-replaceable packs, but if a cell
fails and distorts, try getting it out through the narrow openning
without disassembling the entire unit. I'm not sure how the new
2U units are in this regard.
So although I don't hate the APCs, I'm starting to lean towards
older, standby style units. I suspect that these put fewer failure
points in the power path under normal conditions, which I'm
considering preferable to having the PC fail because the UPS
ran a diagnostic and didn't handle a problem condition correctly.
The UPS shouldn't introduce more problems than it solves.
My $0.02 (CDN)
Kevin B.
Reply to: