Edgar F. ("Ted") Codd

Edgar Frank (Ted) Codd, the youngest of seven children, was born August 19th, 1923, on the Isle of Portland in the county of Dorset on the south coast of England. His father was a leather manufacturer and his mother a schoolteacher. During the 1930s he attended Poole Grammar School in Dorset. He was awarded a full scholarship to Oxford University (Exeter College), where he initially read chemistry (1941 1942). In 1942—despite the fact that he was eligible for a deferment because of his studies— he volunteered for active duty and became a flight lieutenant in the Royal Air Force Coastal Command, flying Sunderlands. After the war he returned to Oxford to complete his studies, switching to mathematics and obtaining his degree in 1948.

As part of his service in the RAF, Codd was sent to the United States for aviation training. That experience led to a lifelong love of recreational flying, also to a recognition that the United States had a great deal to offer for someone of a creative bent like himself. As a consequence, he emigrated to the United States soon after graduating in 1948. After a brief period with Macy’s in New York City, working as a sales clerk in the men’s sportswear department, he found a job as a mathematics lecturer at the University of Tennessee in Knoxville, where he taught for six months.

Codd’s computing career began in 1949, when he joined IBM in New York City as a programming mathematician, developing programs for the Selective Sequence Electronic Calculator (IBM’s first electronic—or at least electromechanical—computer, a huge and noisy vacuum tube machine). He also lived for a brief period in Washington DC, where he worked on IBM’s Card Programmed Electronic Calculator. In the early 1950s, he became involved in the design and development of IBM’s 701 computer. (The 701, originally known as the Defense Calculator, was IBM’s first commercially available computer for scientific processing; it was announced in 1952 and formally unveiled in 1953.)

In 1953, Codd left the United States (and IBM) in protest against Senator Joseph McCarthy’s witch-hunting and moved to Ottawa, Canada, where he ran the data processing department for Computing Devices of Canada Limited (which was involved in the development of the Canadian guided missile program). A chance meeting with his old IBM manager led to his return to the U.S. in 1957, when he rejoined IBM. Now based in Poughkeepsie, New York, he worked on the design of STRETCH (i.e., the IBM 7030, which subsequently led to IBM’s 7090 mainframe technology); in particular, he led the team that developed the world’s first multiprogramming system. (“Multiprogramming” refers to the ability of programs that have been developed independently of one another to execute concurrently. The basic idea is that while one program is waiting for some event to occur, such as the completion of a read or write operation, another program can be allowed to make use of the computer’s central processing unit. Multiprogramming is now standard on essentially all computer systems except for the smallest personal computers.) In 1961, on an IBM scholarship, he moved to Ann Arbor, Michigan, where he attended the University of Michigan and obtained an M.Sc. and Ph.D. in communication sciences (1965). His thesis—which was published by Academic Press in 1968 under the title Cellular Automata—represented a continuation and simplification of von Neumann’s work on self-reproducing automata; in it, Codd showed that the 29 states required by von Neumann’s scheme could be reduced to just eight.

During this period Codd also became a U.S. citizen—though he never lost his British accent, his British sense of humor, or his British love for a good cup of tea.

Codd then returned to IBM Poughkeepsie, where he worked on high level techniques for software specification. He then turned his attention to database issues and in 1968 transferred to the IBM Research Laboratory in San Jose, California. (He subsequently claimed that what initially motivated him in this research was a presentation by a representative from a database company who seemed—incredibly, so far as Codd was concerned—to have no knowledge or understanding of predicate logic.) Several database products did indeed exist at that time; however, they were without exception ad hoc, cumbersome, and difficult to use—they could really only be used by people having highly specialized technical skills—and they rested on no solid theoretical foundation. Codd recognized the need for such a foundation and, applying his knowledge of mathematical logic, was able to provide one by creating the invention with which his name will forever be associated: the relational model of data.

The relational model is widely recognized as one of the great technical achievements of the 20th century. It revolutionized the way databases were perceived; indeed, it transformed the entire database field—which previously consisted of little more than a collection of ad hoc products, proposals, and techniques—into a respectable scientific (and academic) discipline. More specifically, it provided a theoretical framework within which a variety of important database problems could be attacked in a scientific manner. As a consequence, it is no exaggeration to say that essentially all databases in use or under development today are based on Codd’s ideas. Whenever anyone uses an ATM machine, or purchases an airline ticket, or uses a credit card, he or she is effectively relying on Codd’s invention.

Codd described his model further and explored its implications in a series of research papers, staggering in their originality, that he published over the next several years (see annotated bibliography). Throughout this period, he was helpful and supportive to all who approached him—the author of these notes included— with a serious interest in learning more or with a view to helping disseminate, and perhaps elaborate on, his ideas. At the same time, he was steadfast and unyielding in defending those same ideas from adverse criticism.
It should be noted, incidentally, that the relational model was in fact the very first abstract database model to be defined. Thus, Codd not only invented the relational model in particular, he actually invented the data model concept in general.

During the 1970s Codd also explored the possibility of constructing a natural language question and answer application on top of a relational database system, leading a small team that built a prototype of such an application, called Rendezvous. Rendezvous allowed a user with no knowledge of database systems (and with, perhaps, only limited knowledge of the exact content of the database) to engage in a dialog with the system, starting with a query—possibly not very precisely stated—and winding up with a precise query and corresponding answer, where the entire dialog was conducted in natural language (English, in the case of the prototype).

Throughout this time, Codd continued to be employed by IBM. Perhaps because it was heavily invested in its existing nonrelational database product IMS and was anxious to preserve the revenue from that product, however, IBM itself was initially quite unreceptive (not to say hostile) to Codd’s relational ideas. As a consequence, other vendors, including Relational Software Inc. (later renamed Oracle Corporation) and Relational Technology Inc. (later renamed Ingres Corporation), were able to steal a march on IBM and bring products to market well before IBM did. Seeing the way the winds were blowing, senior IBM management decided in the late 1970s that IBM should build a relational product of its own. That decision resulted in the announcement of SQL/DS for the VSE environment in 1981 and DB2 for the MVS environment in 1983.

In Codd’s opinion, however, those IBM products, though clearly superior to their nonrelational predecessors, were less than fully satisfactory because their support for the relational model was incomplete (and in places incorrect). Partly for that reason, Codd resigned from IBM in 1984. After a year or so working as an independent consultant, in 1985 he formed, along with colleagues Sharon Weinberg (later Sharon Codd) and Chris Date, two companies—The Relational Institute and Codd & Date Consulting Group—specializing in all aspects of relational database management, relational database design, and database product evaluation. (C&DCG subsequently grew into a family of related companies, including a parent company called Codd & Date International and a European subsidiary called Codd & Date Limited.)

Over the next several years, Codd saw the relational database industry grow and flourish, to the point where it was—and continues to be—worth many tens of billions of dollars a year (though he himself never benefited directly from that huge financial growth). Throughout that period, and indeed for the remainder of his professional life, he worked tirelessly to encourage vendors to develop fully relational products and to educate users, vendors, and standards organizations regarding the services such a product would provide and why users need such services. He was also interested in the possibility of extending his relational ideas to support complex data analysis, coining the term OLAP (On-Line Analytical Processing) as a convenient label for such activities. At the time of his death, he was investigating the possibility of applying his ideas to the problem of general business automation.

Codd died April 18th, 2003, in Williams Island, Florida. He is survived by his wife Sharon; his first wife Libby; a daughter, Katherine; three sons, Ronald, Frank, and David; and several grandchildren.

Author: C. J. Date