Those who say that computer science is not an inherently dangerous
field have not worked in Wean Hall. What I've learned in my decade
associated with CMU is that Wean's concrete not only looks ugly and
leaks in the rain, but it can also burn and fill with water.
On October 22, 1997, Wean's roof
caught fire. An accident with a propane torch lit the tar on the
roof. It was around that time that we all began to talk about Wean's
fire alarm, a state of the art system circa 1970, that notified
building occupants of problems using a pleasant dinging sound. Picture
a butler calling guests to dinner by politely tapping a fork against a
wine glass - that's the Wean fire alarm. Or, that was the Wean fire
alarm. About a month ago, they finally replaced it with a much more
aggressive system featuring a booming voice telling us all to run for our
lives. But I digress.
By the way, the photo of Wean burning was taken by a web cam at Evolution Systems.
Yesterday a burst water main sent thousands of gallons of water into Wean's third floor data center, where most of the
department's servers were kept. The water came out with such force
that it broke open metal doors and pushed around 300 pound machine
racks like they weighed nothing. This particular water main, about 18 inches in diameter, happened to be located inside my research group's machine room. For years, I'd looked at that particular pipe and thought "better not touch that." I'm proud that my paranoia was so dead on.
This picture is courtesy of Ray Link, who's written about the flood in more detail.
If you're going to put mission critical servers less than 10 feet
from a water main (as we did), it's probably good to have a disaster
response plan (as we did not). This would help cheer you up after you
get the news that the server holding all your source code, papers, and
presentations is under water. In our case, our improvised disaster
response plan is to remove hard drives and place them somewhere warm
to dry out a bit. We also have backups somewhere, I'm told.
Everybody has confidence in the backups. Even so, I think it'll be
some time before 100x100 and
ESM are back online.
This brings me to the poignant part of the story: I am not nearly
as worried about my data as I used to be because on Friday, I told my
advisor I was leaving the program. After nearly 10 years in
Pittsburgh, 6 of them spent in grad school, I'm throwing in the towel.
It's now only slightly dishonest of me to say I've been through hell
and high water, and while that's not the same as saying I've got a
PhD, it'll have to do. I'm quitting grad school while the quitting's
good, moving to New York, and starting a job at Telemetry Investments.