The massive physical and digital search for pioneering computer scientist James Gray came to an end on February 16 after a diverse collection of rescuers decided that there was simply nothing more that could be done. Gray had set out on January 28 on Tenacious, his 40-foot boat, to scatter his mother's ashes near the Farallon Islands, 25 miles off the coast of San Francisco, and has not been heard from since.

The intensity of the rescue effort, in which computer mavens from Microsoft, Google and Amazon wrote software for inspecting satellite imagery, even organizing their own air and water quest apart from those of NASA and the Coast Guard, demonstrated the high regard for the database innovator. Gray's family must now decide if it is time to hold a memorial service.

The esteem for Gray resulted in his receiving, in 1999, the Association for Computer Machinery's A. M. Turing award, considered the top prize in computer science, for his work on databases that led to automatic-teller machines, computerized airline reservations systems and other applications. Gray began his career when computers were still big, slow and difficult to use. He was hired to work on operating systems at IBM's T. J. Watson Research Center in the 1960s.

But, at the behest of a colleague, he decided to switch course to pursue research on the relatively undeveloped software for networking and databases. He quickly ran into difficulties in writing code for IBM's hierarchical database, called the Information Management System, which was originally developed to track inventory for the Apollo space program. A hierarchical database is one in which data are organized in a treelike structure. An employee record for a manager that contains attributes, such as first name, last name and number of employees supervised, would be followed on a lower branch by a record for each of those employees with a similar list of attributes. When using hierarchical systems, Gray sometimes had to let the error-checking "debugging" software run more than a dozen times to determine whether a particular programming statement would execute properly.

Relational databases—in which data are represented in a tabular format—had begun to emerge as an alternative to the hierarchical model. Gray helped foster the emergence of relational databases by programming interfaces that displayed information in these tables in a more visual manner. To promote the handling of multiple requests for data, moreover, Gray and his collaborators had groups of requests processed as an individual "transaction"—the beginning of what is known today as transaction processing, the accomplishment for which Gray is perhaps best known. According to this model, each transaction had to exhibit several properties. All of the actions needed to be processed as a single unit or else they had to be ignored entirely, a characteristic called "atomicity." A transfer from checking to savings should occur as both a debit and a credit, but if the computer crashes while the transaction takes place, it should not be logged at all.

Another property had to do with ensuring that simultaneous transactions are handled cleanly. If a bank customer pays a utility customer online, the utility that receives the remittance cannot witness what is going on until the transaction is finished. In 1980 Gray started designing "fault tolerant" computer systems at Tandem Computer, which were supposed to experience downtime of only one second each century. Although Tandem went out of business, Gray helped devise standards by which high-performance computing is measured.

In 1995 Gray, who had taken up residence as a visiting scholar at the University of California, Berkeley, after a stint at Digital Equipment Corp., went to Microsoft Research. There, Gray headed a team that was building fault-tolerant systems for enormous databases. He worked, along with the U.S. and Russian government as partners, on TerraServer, a Web site that stores more than a terabyte (a trillion bytes) of compressed aerial and satellite photos of Earth. He also was instrumental in building SkyServer, which supplies data from the Sloan Digital Sky Survey, an effort to map a large swath of the universe. And an unmet goal was to build a "World Wide Telescope" that would gather global astronomy databases for access through a single unified interface. Gray's accomplishment of making virtually any amount of data readily available by accessing multiple databases removed limits on how far the Internet can grow.