On Tue, Feb 10, 2009 at 12:40 PM, Jim McAtee <[email protected]> wrote:
> So you purchase two of the same UPS, maintain them as best you can, then the 
> day
> comes when you lose power and _one_ of them fails, what are the chances that
> the other will hold up all of your servers?

  Well, assuming they're not overloaded like they were for us, things
should keep running.  I apparently overestimated the amount of margin
I had on our UPSes.  If I had properly spec'ed things, the servers
would indeed have kept running, despite the suicide of one of the
UPSes.  If I had load tested the configuration after some additional
equipment was added, I would have seen the problem before it became
critical.  This was really due to oversight on my part.  I was all too
willing to believe a number in a management screen, when I should have
done real testing.

  Fundamentally, my problem was a UPS that was overloaded when on
battery but squeaking by on-line.  That problem can happen with a
single UPS just as much.  Indeed, I've seen single UPSes buckle when
they transfer to battery before, too.

  The problem of equipment bought at the same time going bad at the
same time is a real one.  I've seen that across any number of product
categories: Hard disks, printers, whole computers.  But if that
invalidated redundancy, we wouldn't bother with RAID, dual PSUs, etc.

  There's a cost argument, to be sure.  But I'm basically looking at
it this way: For a given runtime, I could use one big UPS, or two
smaller UPSes in parallel.  With one big UPS, a UPS fault means
everything dies.  With two smaller ones, a UPS fault simply means
runtime is reduced.

-- Ben

~ Finally, powerful endpoint security that ISN'T a resource hog! ~
~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/>  ~

Reply via email to