Order Tray | Contact Us | Home | SIG Lists

[aprssig] first.aprs.net

Joel Maslak jmaslak-aprs at antelope.net
Thu Oct 5 00:43:02 UTC 2006


On Oct 4, 2006, at 6:09 PM, John Gleichweit wrote:

> Ouch. You don't have UPS and Generator backup? The server farm for the
> County here has a 750KVA genset and a RFB (really freakin' big) UPS  
> setup
> for the IT Services server farm. Fresno State has multiple server  
> farms and
> each has a UPS and genset of the appropriate size. Heck, the  
> Bulldog ARC
> server farm/radio room/repeater shack has a feed off the UPS and  
> genset.

There's always a single point of failure for a computer system, at  
least that's what I've learned over the years.  Sometimes it might be  
the building's roof or something like that, but no matter what  
precautions you take, sometimes things go wrong - such as generators  
not starting (despite starting perfectly fine for the last 200  
tests), UPSes failing to switch over, transfer switches failing,  
limited supplies of diesel (or oil), cooling systems not on backup,  
lack of a water supply in a power outage, etc.  So it's very possible  
to have backup power and still be severely affected by a power outage.

And you've never experienced life until you've had a smoke detector  
detect a cooling system leak (note that the room looks like it's  
filled with smoke when certain cooling systems leak and you look in  
through a little window), which discharges the fire suppression  
system and properly kills power to the whole room.  Or someone  
accidentally snags the big red button on the way out the door  
(despite it being properly guarded).  Or had the fire marshall tell  
you that the diesel truck can't park there when the generator is  
running.  Or had a generator oil pressure sensor fail 5 minutes after  
the generator starts.  Etc...

Point is there are a lot of things that can go wrong in a modern  
server room - even a properly designed one.  (the examples I gave  
were all examples that really happened to some of the facilities I've  
worked with - these are properly designed data centers with a lot of  
money spent to keep them reliable).  Many of the failures happened  
during a real power outage, despite weekly testing of the power  
systems (need I say that's also a source of failure - what if it  
doesn't work during the test?).

So I'm kind of careful about bragging about backup capability - too  
often the minute you brag about your own, you find out how little you  
have.

It's one of the benefits of APRS, though - it's pretty easy to get a  
station - or even a digi - running without commercial power.  I  
wouldn't count on the phones and internet having the same reliability  
(recently a double cable cut on a major telephone loop in Wyoming cut  
the northern half of the state off from nearly all telecommunications  
- cell phones didn't work, land lines worked only locally, 911 was  
down for dozens of communities, people couldn't buy gas at credit  
card pumps [not a big deal until you realize many Wyoming towns don't  
have attended gas stations due to remoteness], etc - it lasted 8  
hours), as they are a whole lot harder to fix when broken.

With APRS, a D700 parked at the top of the hill is often all the  
infrastructure that is needed.  It won't function like the old  
network did, but it could handle tactical communication for a  
reasonable area easily enough.




More information about the aprssig mailing list