|
FROM THE EDITOR
Edgar (Ted) Codd, 1923-2003
Remembering the father of the relational database
By all accounts, Edgar F. Codd, known as Ted, was a brilliant man. Among his accomplishments was the development in the early 1970s of a relational model for data management—a sophisticated and complete theory for storing and manipulating large amounts of business data. Relational databases built according to Codd's design form the foundation of business today; banks rely on them to track the flow of capital; retailers use them to monitor inventory levels; HR departments use them to manage employee accounts; libraries, hospitals, and governments store millions of records on them; in fact, nearly every business in the world uses a relational database in some capacity. And during the 30 years since Codd made his theories public, relational databases have become a nearly US$13 billion per year industry.
Early Life
Ted Codd was born in 1923 to a large family in Portland, Dorset, England. He attended Oxford University, where he majored in mathematics and chemistry, and then served in the Royal Air Force during World War II. After the war, Codd set off for New York and landed a position as a mathematical programmer with IBM. Codd's first project was to help build an early computer called the Selective Sequence Electronic Calculator, said to take up two floors of a downtown office building.
After earning a doctorate in computer science at the University of Michigan in the mid 1960s, Codd transferred to IBM's development laboratories in San Jose, California, where he began working on the relational model of data management, a model that relied heavily on mathematics.
Improving the Database
Early computers were too large and expensive to have broad application in business. In the 1960s, they became cost-effective and widely adopted by the private sector, and standards and languages were developed specifically for business use. Among these were two models for working with data, a
hierarchical and a related network model.
In the hierarchical model, data records relate to one another in a hierarchy; a main record is at the top layer, and subsequent types of records branch below. In the network model, a set of records in one layer could potentially belong to two different containing hierarchies at the next upper layer. For both models, writing queries to retrieve information required a deep understanding of the navigational structure of the data itself—a complicated task generally left to expert programmers.
Codd proposed a new solution. In a series of reports culminating in the 1970 groundbreaking technical paper, "A Relational Model of Data for Large Shared Data Banks," Codd suggested that data be stored independently from hardware and that a programmer use a nonprocedural language for accessing data. The crux of Codd's solution was that data, rather than being stored in a hierarchical structure, be stored in simple tables composed of rows and columns in which columns of like data would relate tables to one another. A database user or application, in Codd's way of thinking, would not need to know the structure of the data in order to query that data. Soon after publishing the paper, Codd issued even more-detailed guidelines, releasing his 12 principles governing the creation of relational databases. (That set of principles would grow to the hundreds by the late 1990s.)
After Codd's theories were published, they were not immediately adopted by IBM. The company had invested heavily in a hierarchical database system called IMS, and it was left to other companies and entrepreneurs to figure out how to further develop Codd's ideas. Chief among them was Larry Ellison, who along with Ed Oates and Bob Miner produced the world's first commercially available relational database management system in 1977, and in the process, started the company that would become Oracle. The rest is database history.
But history didn't stop there for Ted Codd. Although Codd remained affiliated with IBM through the early 1980s, he also started a consulting company with longtime collaborator Chris Date, and continued to research and publish papers on topics such as data normalization, analytics, and data modeling until his death earlier this year.
Jeff Spicer, Editor in Chief
jeff.spicer@oracle.com
| 
Bookmark
