System R was the experimental database system in which IBM set out to prove that Edgar Codd’s relational model was not just elegant theory but a workable basis for a real product. The foundational paper, “System R: Relational Approach to Database Management” by Astrahan and a large team at IBM San Jose (ACM Transactions on Database Systems, 1976), describes a system that “provides a high level relational data interface” and gives users “a high level of data independence by isolating the end user as much as possible from the underlying storage structures.”
The same paper lists the data-control machinery that made System R more than a query toy: “authorization, integrity assertions, triggered transactions, a logging and recovery subsystem, and facilities for maintaining data consistency in a shared-update environment.” Many of these ideas, including transactions, logging, and recovery, became standard features of every serious relational database that followed.
Three innovations from this project shaped the entire industry. System R introduced the query language SEQUEL, later renamed SQL, as a high-level, declarative way to ask for data. It introduced a cost-based query optimizer that chose execution strategies automatically rather than forcing the programmer to specify them. And it demonstrated transaction processing with the consistency guarantees later summarized as ACID.
System R was a research prototype, not a shipped product, but its design flowed directly into IBM’s commercial relational databases. IBM Research lists the work as a landmark publication, and the SQL/DS and Db2 products that IBM brought to market in the early 1980s grew out of the System R effort.