Now, companies are trying to keep improving when profits aren't gushing. "The exploration and production industry is very technically savvy," says Richard Ross, VP and CIO of Amerada Hess Corp., a $13.6 billion oil and gas company in New York. Amerada last year saved several million dollars by moving a building-sized oil rig in Algeria in seven days instead of 11, thanks to a knowledge-management system that pooled the know-how of experts via the Web. Low oil prices mean "we may move fewer rigs," Ross says. "But if we can do it in four days less, we'll save money."

More CIO Insights

White Papers

• How Executives and Boards Can Help Improve Large Technology Projects
• BSM: What It Is and Why You Should Care

Reports

• Can IT Save Our Schools?
• The Open Source Enterprise

Videos

She considers business acumen just as important as technical knowledge for a CIO. Here's why.

But the recession and aftershocks of the Sept. 11 attacks have deflated oil prices, and oil-company stocks fell last year after big gains in 1999 and 2000. That puts pressure on budgets, including IT. West Texas Intermediate closed at $24.16 per barrel on March 13, compared with more than $32 on that date in 2000; natural-gas spot prices this year are projected to average less than half of last year's, though the government and investors predict a rebound next year.

The U.S. petroleum industry spends about $9 billion annually on IT, says oil industry analyst Randall Nottingham at Strategy Analytics. That's not counting spending on drilling technologies and robotics that don't show up in IT budgets (see story, p. 78). But budgets are being crimped just as the industry faces a heavy round of merger-related integration work and a looming shortage of professional talent that requires CIOs to respond with new ideas. Topping their lists: assigning scientific computing jobs to cheap business data servers, outsourcing more bulk data processing to specialists, investing in Internet technologies to decentralize massive storehouses of data and to transmit organizational knowledge, and building Web portals to share data with partners.

The challenge for IT managers at the major petroleum companies, midtier independents, and oil-field services companies in the so-called upstream exploration and production market is to process the petabytes of data afforded them by the rapid improvements in data-collection techniques. "Just like the high end of any market, there will always be pressure for more, faster computing cycles," Amerada Hess CIO Ross says.

The huge cost of finding and extracting oil and gas--which can run into hundreds of millions of dollars for offshore fields and more than $1 billion for deep-water projects a mile beneath the ocean's surface--means competing companies often form joint ventures to spread financial risk. This requires IT departments to share sensitive exploration information while protecting other assets. A rush of consolidation that in the past two years has wed Exxon and Mobil, Chevron and Texaco, and BP Amoco and Atlantic Richfield is exacerbating the problem of making terabytes of geophysical, engineering, and business data, stored largely on tape, available to dispersed workers.

Making matters trickier is a projected shortage of geologists, geophysicists, and petroleum engineers stemming from an aging workforce. It's why companies are experimenting with knowledge-management applications built on E-mail systems, Web portals, and interactive workspaces in hopes that their expertise won't retire with their workers.

As in other data-and computationally intensive businesses, the technology strategies of upstream petroleum company CIOs could ripple though the IT market. Large IT vendors such as Compaq, Dell Computer, IBM, and Sun Microsystems say the needs of upstream petroleum companies guide their decisions in areas such as clustering technologies, high-performance computing, Linux systems, data integration, and microprocessor platforms.

"Oil companies are probably second to NASA in data volumes," says Mike Heagney, Sun's global energy manager. "The systems still aren't fast enough; they consume pretty much anything we put out." That includes a 106-CPU symmetric multiprocessing machine that oil companies use to analyze seismic data produced by sending vibrations through the ground via explosives or machinery. The speed and pattern of the vibrations yield clues about geologic structures--folds in the earth's surface--that may yield oil.

Technical advances let geophysicists collect seismic data with greater sensitivity, and exponential improvements in microprocessing power and data storage let scientists visualize the data in finer detail. The result: Scientists are more certain not only of the presence of petroleum-yielding sandstone and limestone, but of whether oil and gas are likely to seep away from the strike point. In a business in which only about 30% of the oil in a reservoir can be extracted, even small improvements in accuracy can be a competitive edge.

All that improvement also inflates the amount of seismic data companies have to handle. Not surprisingly, collecting digital representations of sound waves generated by rocking the earth's surface with explosives is a noisy process. Exploration scientists have to employ CPU-chewing algorithms to clean up the data, which adds to storage volumes and processing requirements. Amerada Hess' Houston data center houses 100 terabytes of seismic, financial, and well-log data--gathered by measuring properties of materials that drills bore through--on tape and disk. "It's not an efficient process to move it around," senior systems programmer Jeff Davis says.

To hold down costs and better distribute information to employees, Amerada Hess, like others in the market, uses clusters of Intel servers running Linux, linked by Ethernet connections, to handle parallel data-processing chores previously done on an IBM RS/6000 SP supercomputer. "That gives us cheap, scalable power wherever we need it," says Ross, who estimates his company's 200-node Dell cluster runs jobs for about $100,000 a month--10% the cost of the SP. "We're looking for ways to manage data on a more distributed basis so we don't have to ship it all the time," he says. "On the technical computing side, we're seeing more of our apps being ported to Linux, which I'd imagine would be very threatening to vendors like Sun."

Sun's not sitting still. It's scaling up its high-performance computing capabilities to keep pace with seismic processing requirements. In November, Sun began work with the University of Houston to develop a 104-CPU cluster that employs "grid computing" software to assign underutilized computing resources to jobs that need them. One top application: seismic imaging. "In some markets, it looks like Moore's Law isn't enough," says Fred Kohout, director of marketing for Sun's technical markets group.

	The upstream sector is the most tech-savvy, IBM's Painter says.

Meanwhile, IBM is making hay in the oil-exploration market. It's sold Linux clusters to customers such as Royal Dutch/Shell Group, ChevronTexaco, No. 2 oil-field services company Schlumberger's WesternGeco unit, and French seismographic services concern Compagnie Générale de Géophysique. "This is the largest industrial technology market that exists," says Don Painter, a petroleum industry executive at IBM Global Services. "Within that, the upstream market is the most technically innovative."

Dell, which also claims WesternGeco and CGG as customers, is trying to increase its presence in the high-performance computing market. Last month, the vendor began selling preconfigured clusters of its PowerEdge servers aimed at scientific, engineering, and financial-modeling apps used by petroleum companies and other customers. Two Dell clusters at CGG are listed on Top500.org, an online ranking of the largest supercomputer installations maintained by the University of Mannheim in Germany and the University of Tennessee. Also last month, Dell struck a three-year deal with supercomputer maker Cray Inc. to resell Dell machines that can be clustered in high-speed networks. Cray scientists will provide assembly, setup, and maintenance for the clusters. "Petroleum companies are buying more Intel for exploration. It's a total-cost-of-ownership issue," says Linda Callahan, executive director of the Cornell Theory Center, a high-performance computing research facility that advised Dell on its program.

Running jobs on small Linux servers means WesternGeco can perform some data processing onboard surveying ships, saving time, Venkataraman says.

But not all upstream-computing jobs run well on Linux clusters. WesternGeco's "bread-and-butter processing" of geological surveys of Alaska and the Gulf of Mexico, for example, is still done on IBM SP supercomputers, says Kannan Venkataraman, area manager for worldwide computer systems at WesternGeco, which gathers and analyzes seismic data for clients.

As oil companies outsource more seismic processing, especially the commodity depth-migration algorithms that smooth out noisy data, low bids and quick turnaround are king at companies such as WesternGeco, Halliburton, CGG, and Veritas DGC. That's where Linux clusters come in--they've reduced processing costs more than tenfold, Venkataraman says. "You don't want to tie up expensive computing cycles or storage on stuff they don't need to do," he says.

About a year ago, WesternGeco ported its seismic software suite to Linux. The repetitive nature of depth-migration jobs means they can be broken down into parallel chunks that run on Western's 512-node Dell and IBM Linux cluster for days or weeks at a time. Running jobs on small, off-the-shelf Linux servers also means the company can perform some data processing onboard its surveying ships, saving time.

Still, not all petroleum companies are fans of IT outsourcing. Yes, they can save money by moving process-intensive, undifferentiated work off their balance sheets. But it also means that Schlumberger and Halliburton's Landmark Graphics Corp. IT subsidiary conduct a larger share of advanced research and development on upstream-petroleum technologies. That diffuses knowledge throughout the industry, making competitive advantages short-lived, says an IT manager at one midsize U.S. petroleum company, speaking on condition of anonymity. "Companies try to get an edge on the computational side," he says. But "it's a really fleeting amount of time you can maintain that edge before Schlumberger and Halliburton sell it to someone else."

Joint ventures between petroleum companies, or between companies and governments that own reservoirs, also tend to spread expertise. Contractually, the operator of a joint venture, which supplies the manpower, needs to share enough data with its partner that both parties can decide whether to improve the site and where to drill. IT shops must balance the need to share more information electronically, while guarding trade secrets.

"At any given time, our partners are our competitors," Amerada Hess' Ross says. "We need to share knowledge and still protect our assets. The Internet is an important tool to do that." The company is considering creating "community spaces" for partners inside its firewall, in which users can see seismic, production, and accounting data, while shielding other information via proxy server software.

Shell is conducting a pilot with the British government to cut down paperwork by filing regulatory reports via the Internet. Neil Carmichael, a member of Shell's exploration and production E-business team, says electronic collaboration between companies and governments could increase as the nonprofit Petrochemical Open Software Corp. defines XML standards for the upstream business.

Meanwhile, workforces are becoming more dispersed. Since the 1980s, oil companies have shed departmental structures in favor of virtual teams of geophysicists, engineers, accountants, and other staff who work on groups of oil and gas assets. Now, in addition to mergers, acquisitions, and joint ventures, many in the business expect a shortage of trained North American petroleum scientists and engineers. That could lead to more global hiring.

Bill Severns, an analyst at Cambridge Energy Research Associates, estimates the median age of these "petro-professionals" at 47 in North America and Western Europe. As they retire, "in five to 10 years, there might not be as many people as you'd like working in North America."

Blame in part the oil superboom of the late '70s and early '80s, when oil prices spiked. Petroleum companies went on a hiring binge of technical staff, then held down hiring when oil prices collapsed in 1986 and kept staff levels lean during a 15-year period of productivity gains. "Experienced petro-professionals have built up a tremendous amount of knowledge and expertise," Severns says. Now, oil companies are hiring staff from Russia, China, India, and elsewhere. "How do you transfer that knowledge and expertise to this new generation without disrupting performance?" That's just one of the many quandaries the oil exploration and production industry will look to IT to help solve.