North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Power monitoring Re: Power Outage in Chicago Loop

  • From: Sean Donelan
  • Date: Mon Oct 09 08:30:35 2000

On Sun, 08 October 2000, Kevin Day wrote:
> I'm currently about 30 miles north of the loop... at 12:46 the voltage
> coming into the building here was 140-160 volts for several minutes(according to
> our UPS). I'm wondering if the power was just going nuts everywhere else,
> causing ameritech equipment to go nuts. Possible explaination for packet
> loss?

After my first summer in PG&E country, I've been wondering if there was
a way for ISPs to share power quality data about the local utility.  For
the most part, every ISP in a region experiences the same woes and problems
of the electric utility. Most ISPs are capable of at least minimal monitoring. 
If the shared data was limited to only the upstream side of the ISPs power
system, it would show the performance of the utility; but ISPs could still
keep any internal problems secret.  While a power quality meter would be
nice, even SNMP capable UPSes can report basic data.

One of the reasons I'm interested is I've needed to research what things
like the CBEMA curve really means, and discovered how little hard data
exists about power and computer centers.  Almost all the data comes from
three studies in the 1970's by IBM, Bell Labs and the US Navy.  Since then
we've transitioned from mainframes to minicomputers to pc's; and the power
supply industry is undergoing deregulation.

Like most industries, there are academics who would love to crunch any
data they could get. The question is how much data could we get.

What's in it for ISPs?

Essentially the same things as when IBM, Bell Labs and the US Navy did
their studies two decades ago.

What should we design our systems to handle?  Are things really as
bad as the UPS and surge protection vendors lead us to believe?

What is "normal" power throughout the country? How severe can power get?