INTEGRATION MAKES INDUSTRY SEEK WAYS TO BREAK DATA BOTTLENECK

Allen Johnson
Convex Computer Corp.
Dallas

Technologies such as 3D seismic, subsurface imaging, and cross-disciplinary data integration have transformed exploration and production.

Today's multidisciplinary teams prove reserves that couldn't be detected with techniques of the 1970s or early 1980s.

As these information technologies evolve they spawn challenges. Chief among them is data management.

Studies of data facilities show that, in many cases, geoscience professionals spend as much as 80% of their time on tedious data-management tasks such as locating, retrieving, and preparing information. They thus spend only 20% of their time on analysis and interpretation.1

Multi-terabyte (trillion-byte) systems are nearly ready to break the data bottleneck at affordable cost. Oil finders in the late 1990s will be able to access the largest data sets by pointing and clicking on a map display of their area of interest in an arrangement called an intuitive data browser.

The browser will let professionals focus on geoscience rather than on juggling data. It's a simple concept. But its implementation stretches the limits of what today's hardware and software can accomplish and exceeds the capabilities of any single vendor.

As oil companies' data management requirements escalate, tightly coordinated vendor teams are delivering the first generation of workable multi-terabyte solutions. The next generation, based on emerging technology and offering a cost-effective growth path from current solutions, is expected to enter the market in 2-3 years.

What are the current and future solutions? What do they look like? How will tomorrow's solutions support intuitive browsers? This article discusses the data management challenge, its hidden costs, and the strategic importance of properly scaled solutions.

SCALING THE CHALLENGE

Consider what it takes to manage geodata. A 3D survey covering one Gulf of Mexico lease block roughly 10 sq miles and including a fringe area may generate 50-100 gigabytes (billion bytes) of data, which must be stored and processed.

In recent years companies have been acquiring 3D seismic on increasing scale. A megasurvey can now cover 100 blocks and produce 10 terabytes of 3D data at a cost approaching $100 million.

Moving, storing, and retrieving gigabyte to terabyte chunks of data pose a major challenge. For example, storing a single 100 block marine 3D survey on 9 track tapes would require 63,000 tapes costing over $1 million, and 5,000 sq ft of climate-controlled storage.

Other forces are driving up the amount of geodata used by project teams. E&P teams increasingly seek to analyze raw 3D seismic, for instance using amplitude vs. offset (AVO) analysis to determine if bright spots indicate natural gas, are wet, or exist merely as artifacts of the acquisition process. But raw seismic data sets are typically 50-100 times the size of the processed seismic, magnifying data retrieval and delivery problems.

At the same time, 3D seismic has become a basic requirement for financing and developing many onshore prospects. As a result, 3D land surveys, though generally smaller than marine surveys, are proliferating. And the trend toward multidisciplinary teams has increased the need for integrating 3D seismic with well logs and production data, further escalating the use of large data sets.

More data, larger data sets, and more frequent usage are trends likely to continue, even intensify. So are data bottlenecks. E&P companies thus seek solutions on a scale that can accommodate tomorrow's geodata needs.

AN EXAMPLE

Modular, multi-terabyte solutions can dramatically affect data management and provide a cost-effective growth path to next generation data management technologies.

A major oil company was faced with storing and managing terabytes of data acquired during 30 years of exploration and production. The company gradually developed a data management solution built around a super fileserver computer.

The superserver can pump data out of storage at sustained rates approaching 20 megabytes/sec, fast enough to keep pace with the needs of the company's compute-intensive machines and workstation-using E&P teams. High storage capacity makes multi-terabyte, on-line data access cost-effective for the first time.

The superserver is integrated with automated storage and high-speed input/output facilities, which also can retrieve and send data at 20 megabytes/sec. The solution includes large storage management software and a graphical user interface developed by the oil company.

Table 1 lists some of the major costs that were identified and compares the company's prior and current data management solutions. Not quantified are hidden benefits such as data security. Because the company's prior manual system was relatively costly and cumbersome to operate, it was impractical and costprohibitive to make backup copies of much of the archived data.

The current solution's relatively low cost per gigabyte, speed, and ease of use enable the company to cost-effectively back up more of its data, substantially improving data security.

Also not quantified is the company's retrieval cost per data request. With the prior manual system, this included the time required to walk the aisles, collect tapes, and carry them to a staging area. Retrievals that formerly took hours or days are handled by the automated system in minutes or seconds.

Retrieving data quickly and delivering it on demand to seismic processors and project teams aren't data management's only missions but are by far the most crucial. By improving productivity, they help boost the company's return on investment (ROI) in compute-intensive machines, project teams, and stored data. More important, when geoscientists and engineers are free to focus on the project at hand rather than on handling data, they have time to produce better information.

GROWTH PATH

A few years from now, what will geodata management look like?

Solutions will include today's basic elements: an automated storage and retrieval system tightly integrated with some type of fileserver (the data delivery engine); a high speed network; data management software; and a graphical user interface (Fig. 1).

The storage/retrieval component will resemble today's large-scale systems: a robotic picker, high speed tape drives, and racks of tape cassettes-currently high capacity tapes, each able to store 15-165 gigabytes.

However, the data delivery component may look very different. A technology called scalable parallel processing (SPP) is emerging. Designed to accommodate very fast throughput, SPP may become the basis of a new data management architecture. Once proven, it could eliminate the need for a separate data delivery engine.

Data delivery and intensive computing could be combined in one machine (Fig. 2), making geodata management potentially more efficient.

Does that mean today's investment in a high-speed storage/retrieval system and superserver would be wasted? The reverse is true. Today's high-speed, multi-terabyte solutions employ essentially the same physical storage/retrieval that tomorrow's will use. Such storage systems represent more than half the cost of the integrated storage/server combination. The high-speed server could be changed out cost-effectively in an operation that would preserve investment while accommodating data growth and rapidly evolving server technology.

INTUITIVE DATA ACCESS

Data should be readily accessible in ways that make sense to map-oriented professionals. This means a geodata management system should provide a map-oriented, graphical, user interface - an intuitive data browser.

Such a system prevents E&P professionals' having to deal with the complexities of the data management infrastructure.

User interfaces of this type have been developed by independent vendors. One intuitive browser, designed for use with large storage systems, will enable the E&P professional to bring up a map display, select an area of the map and specify various data attributes from layered pop-up menus (Fig. 3).

Integrated vendor teams generally will offer high-speed, multi-terabyte systems as a custom solution. The data management infrastructure, hidden behind the intuitive interface, will know what data to retrieve because the oil company's underlying data sets will be cross-referenced. 3D survey data, for example, will be cross-referenced by seismic tape number with shot-point map, scanned image location for the stacked seismic section, and observer's logs corresponding to the shot-point ranges.

Different data bases and data types also will be cross-referenced. This will enable the system to respond when users make requests by attribute-well number, stratigraphic top, core data, seismic line and shot point, seismic horizon, and velocity. Well logs, drilling reports, and geologists' reports will be referenced and readily available on the system.

The intuitive browser and underlying data infrastructure will enable geoscientists and engineers to quickly obtain meaningful, integrated information on an ad hoc basis. By applying all of a project's available data, E&P teams can deliver better analyses and interpretations.

Comprehensive data management solutions including high speed networks and application software interfaces will make the results readily available to project work groups, corporate managers, and industry partners.

BREAKING THE BOTTLENECK

Over the next few years, geodata growth will continue. Large E&P companies need comprehensive systems to store the data, rapidly retrieve and deliver it to users, and make the data easily accessible.

Tightly integrated vendor teams are beginning to provide solutions that offer cost-effective, scalable, terabyte-range storage plus fast data delivery, data management software, a map-oriented user interface, and a cost-effective growth path to next-generation technology.

By giving E&P teams more time to focus on their mission, comprehensive data management solutions break the data bottleneck, enabling E&P companies to increase returns on investment and improve decision support throughout the organization.