Case Studies
The Situation
With a goal of replacing 100 percent of its petroleum
reserves annually, HESS directs more than 95
percent of its total capital expenditures to
exploration, development and production activities.
After evaluating hundreds of opportunities, the
company drills approximately 15 oil wells per year.
Traditionally a high-stakes guessing game, oil
exploration today uses technology to improve the
odds of success. At its Houston, Texas, offices, HESS
maintains a 20,000-square-foot data center servicing
the s
eismic processing needs of its global exploration
and production operation. This data center processes
huge amounts of raw seismic data, collected
worldwide, to help determine the presence and
location of oil deposits. Creating the three-dimensional
models used by geoscientists to find oil and gas in
hard-to-reach underwater areas requires massive
computing power.
A
t an
y gi
ven time, the data center processes
1,500 Terabytes (TB) of active data via Linux clusters
operating on blade-server arrays. The company’s
processing needs have grown by 25 percent per year
for the past few years, and managers expect similar
gr
o
wth in the years to come.
The data center’s computer hardware evolved rapidly
from large mainframes only a few years ago to today's
d
is
tr
ib
uted arrays of 1,000 blade servers. This increased
computing capacity supports more precise modeling
o
f geologi
cal f
ormations and increased accuracy in
identifying potential oil and gas deposits. The
continued evolution of the hardware requires an
eq
u
all
y d
yn
ami
c support infrastructure.
“I always have to be prepared for change,” says Michael
Mu
s
gr
o
ve, supervisor of Computer Operations for
HESS’s Exploration & Production division. “We could
decide tomorrow that we need to add another
1,000-server array.”
A
s with m
an
y d
a
t
a c
e
nter managers, Musgrove was
ca
ught somewhat off guard by blade servers’ heat
output. “The vendors of these high-density servers
don’t always tell you up front how much heat they
r
e
all
y ki
ck o
ut and how much power they draw,”
he says. “It’s been a learning curve.”
Whe
n HE
SS installed its first supercomputing cluster,
Musgrove attempted to turbo-charge his existing
raised floor cooling systems. “We installed Plexiglas
A Customer Success
from the Experts in
B
usiness-Critical Continuity™
“People often underestimate the power and cooling
demands of high-density equipment. Those 1,000
blade servers draw 360 kW of power and put out
104 tons of heat in 2,000 square feet.”
Michael Musgrove, supervisor of Computer Operations,
HESS’s Exploration & Production Division