Main Page | See live article | Alphabetical index


This article is about a relational database system. For the artist, see Jean Auguste Dominique Ingres.

Ingres was an early relational database system, created as a research project at the University of California, Berkeley starting in the early 1970s and ending in the early 1980s. The code, like that from other projects at Berkeley, was available at minimal cost under a version of the BSD license. By the mid-1980s Ingres had spawned a number of commercial database applications, including Sybase, SQL Server, NonStop SQL, Informix and a number of others. A follow-on project started in the mid-1980s as Postgres, leading to the development of PostgreSQL, Illustra, and later versions of Informix. By any measure, Ingres is one of the most influential modern computer research projects.



In 1973 when the System R project was getting started at IBM, the research team released a series of papers describing the system they were building. Two scientists at Berkeley, Michael Stonebraker and Eugene Wong, became interested in concept after reading the papers, and decided to start a relational database research project of their own.

They had already raised money for researching a geographic database system for Berkeley's economics group, which they called Ingres, for Interactive Graphics REtrieval System. They decided to use this money for to fund their relational project instead, and used this as a seed for a new and much larger project. For further funding Stonebraker approached the DARPA, the obvious funding source for computing research and development at the time, but both DARPA and the Office of Naval Research (ONR) turned them down as they were already funding database research elsewhere. Stonebraker then introduced his idea to other agencies, and, with help from his colleagues he eventually obtained modest support from the NSF and three military agencies: the Air Force Office of Scientific Research, the Army Research Office, and the Navy Electronic Systems Command.

Thus funded, Ingres was developed during the mid-1970s by a rotating team of students and staff. Ingres went through an evolution similar to that of System R, with an early prototype in 1974 followed by major revisions to make the code maintainable. Ingres was then disseminated to a small user community, and project members rewrote the prototype repeatedly to incorporate accumulated experience, feedback from users, and new ideas. Ingres remained largely similar to IBM's System R in concept, but based on "low end" systems, Unix on DEC.


Unlike System R, the Ingres source code was freely available (on tape) for a modest fee. At the same time DEC minicomputers were in the process of taking over much of the educational computing market. By 1980 some 1,000 copies had been distributed, and a number of companies were using the code for their own product lines.

Informix was one of the earliest users, and one of the few that was formed by people completely external to the Ingres project. They released the first version of their Ingres-based product in 1984, and by 1997 had become the #2 database vendor. However a series of management and accounting blunders destroyed the company's credibility in two short years, and in 2000 IBM purchased the remains.

Robert Epstein, the main programmer on the project while at Berkeley, formed Britton-Lee, and then Sybase. Sybase had been the #2 product (behind the System R based Oracle) for some time through the 1980s and into the 1990s, before Informix came "out of nowhere" and took over in 1997. Sybase's product line had also been licensed to Microsoft in 1992, who re-branded it as SQL Server. This relationship soured in the late 1990s, and today SQL Server outsells Sybase by a wide margin.

Jerry Held and Carol Youseffi moved to Tandem Computers where they built a system that evolved into NonStop SQL. NonStop was a version of Ingres modified to run effectively on parallel computers, adding functionality for distributed data, distributed execution, and distributed transactions (the last being fairly difficult). First released in 1987, a second version in 1989 added the ability to run queries in parallel, and the product became fairly famous for being one of the few systems that scales almost linearly with the number of processors in the machine: adding a second CPU to an existing NonStop SQL server will almost exactly double its performance. Tandem was later purchased by Compaq who started a re-write in 2000, and now the product is at HP.

Eventually Stonebraker himself left Berkeley to found Ingres Corporation in 1982 to commercialize the product. This version was soon purchased by Computer Associates, who continue to offer it today. Stonebraker later moved along to Sybase, Informix and Illustra (see below) before returing to Berkeley in 1985.


After returning to Berkeley, Stonebraker started a post-Ingres project to address the problems with the relational database model that had become increasing clear during the early 1980s. Primary among these was the relational model's inability to understand "types", combination of simpler data that make up a single unit. Today we typically refer to these as objects.

The resulting project, named Postgres, aimed at introducing the minimum number of features need to add complete types support. These included the ability to define types, but also the ability to fully describe relationships – which up until this time had been widely used by maintained entirely by the user. In Prosgres the database "understood" relationships, and could retrieve information in related tables in a natural way using rules. For complete details, see the PostgreSQL article.


(needs work)

Besides its being truly relational, another QUEL advantage over SQL is that it is based on relational calculus, while SQL is based on an inconsistent mixture of versions of both relational calculus and relational algebra.


The Rise of Relational Databases
The History of PostgreSQL Development