Grid: Seamless Computing for Maximum Power

In the last two articles, I introduced two concepts that help standardize software architecture complexity and interoperability. Web services standards allow developers to build software components that interact with other components in a predictable manner. Services-oriented architectures provide a standardized framework within which components may interact and be combined into workflows. Distributed computing is the next step in complexity and capability. Distributed computing describes networked computers that are used in tandem to solve problems, provide functionality, or store data. Grid computing is the most recognizable technology that has emerged from industry experiments with distributed computing. Grid computing encompasses the concept of seamlessly sharing applications, data, storage, processing power, and hardware across a dynamic network that may be distributed over a broad geographic area.

The most successful example of distributed computing that foreshadows the power of true grid architectures is the SETI@home project. Participants in SETI@home, usually computer hobbyists and space junkies, can download a screensaver application that uses their computer to search for intelligent signals within cosmic noise when the screensaver is invoked. To date, SETI@home has had 5,306,848 unique participants, processed 1.7 billion results, taken advantage of more than 2.1 million years of otherwise unused computer time, and has performed 6x1021 floating point operations. That means that in the 10 years since the conception of the project at a Christmas party, SETI@home has performed approximately 1 hundred million operations per second. That is the power of a single-function, closed application architecture written in C that simply takes advantage of your sleeping computer!

A research group at the University of Leeds developed a spatial decision support system using Sun Grid Engine technology. The Hydra application uses distributed databases and computing tools to identify appropriate networks of health facilities for patients based on age, sex, treatment type, and other pathological criteria. In the future, applications such as Hydra may allow users at different locations to seamlessly access broad quantities of data and functionality in Grid-based SSDS.
SETI@home is a relatively simple cousin of a true grid architecture. Actual grid systems require gridded computers to share data and processing and cannot be dependent upon the whims of home computer users; therefore they are more complex and have taken longer to evolve. Sun Microsystems, Inc. took a relatively early stab at providing an application development framework for distributed applications with JINI technology. JINI, in the term used by Sun developers, "federalizes" numerous distributed computers, allowing them to be administered and to operate as if they are one large system. Sun also started the open source Grid Engine project to further grid technology and now sells one of the fruits of the project as a packaged software application. Oracle Corporation made the most recent recognizable entry into grid space with its 10g database product. 10g attempts to allow the seamless distribution of data storage and database processing capability across numerous servers in diverse networked geographic locations. The grid capabilities of 10g are reported to include the capacity to administer thousands of servers from one location, the ability to store and access more than 100 petabytes of information, and the security facility to allow a single user login to access all of the 10g grid instance's capability. Several other companies and projects are further exploring the development of grid computing.
Grid is such a nascent technology that geospatial applications examples are currently few and elusive. The state of New Mexico Department of Transportation has been reported to be designing a system based on grid technology that will store extremely high resolution aerial photography for every square foot of its highway system. A group at the University of Leeds in the United Kingdom has used grid technology to build a spatial decision support system (SDSS) to solve health care decision problems. Given the fact that geographic problems often use massive quantities of data, consume immense processing power, and typically require complex hierarchical or network computations, grid seems like an ideal architecture for complex geospatial computing problems. Currently, grid is a solution looking for more geospatial problems. GIS software vendors and open source projects should be examining the power of grid as a vehicle to facilitate the expansion of GIS technology into new and more data-rich markets.
Acknowledgments

The author would like to thank Dr. Mark Birkin of the School of Computing and the School of Geography at the University of Leeds for information and graphics.