SQL/DS: IBM's First RDBMS

In the late 1970s, IBM software labs were aligned with the IBM hardware families. The decisions to commercialize the relational database prototype called System R, which had been developed during the 1970s at the IBM Research facility in San Jose, California, were made based on a hardware family business case. The Endicott Lab, supporting the small- to mid-sized mainframe environments running VM and VSE, had the skills and the competitive pressure to launch the relational database management system (RDBMS) commercialization project in 1979, and it delivered SQL/DS two years later. This article traces how SQL/DS, running on VSE and then on VM, became IBM's first commercial relational database in 1982, over a year before the availability of DB2 running on MVS.

Focused on Hardware

Until the mid-1980s, IBM was largely organized around its hardware families of mainframe, midrange, and personal computers. System software was developed by different labs within these hardware-based organizations to exploit the capabilities of each hardware platform and to provide vertically optimized solutions that would drive their use. The needed software was mostly hardware specific, such as operating systems, device controllers, and file systems. When relational database technology emerged from IBM's San Jose Research facility through the System R research project, IBM certainly didn't see it as a new software business category, but primarily as a way to further increase demand for hardware. Around 1979, the decisions to exploit this new opportunity were independently evaluated by the hardware groups, for each operating environment, based on strategic initiatives and market-specific pressures.

In 1979, MVS was the premier IBM operating environment, with a reputation for high performance, quality, and high availability. These attributes were strengthened by a lengthy field trial process for new releases. IMS, the popular online hierarchical database and transaction-processing system on MVS, along with nonrelational DBMS products from independent software vendors (ISVs) were successfully driving mainframe growth, and the platform had little competition. Customers desiring compatibility and consistency between database usage on MVS and VSE chose the Customer Information Control System (CICS) as a transaction manager, which could operate with IMS databases on MVS or DL/I databases on VSE (or even the database systems from the ISVs).

Thus, introducing any new database offering based on System R required careful positioning. An MVS product would require a significant rewrite from the VM version of System R. It would also need to be positioned relative to the already successful IMS product to help customers understand which to use. Finally, it would need extensive quality and performance testing along with lengthy field trails to not introduce any weakness into the fastest growing mainframe platform. An MVS version of System R would not be produced quickly.

VSE, on the other hand, was used as a transaction-processing environment by thousands of mid-sized customers, but without the widespread use of databases. Mission-critical, real-time VSE applications usually accessed data files through CICS or with non-IBM transaction-processing offerings. There was limited use of DL/I and other nonrelational databases. VM was used primarily as a development environment or hosting environment for VSE. A commercial version of System R could be introduced into VSE or VM more quickly than into MVS, with less development effort and product positioning.

Internal Competition in IBM

The IBM Santa Teresa Lab (STL, now called the IBM Silicon Valley Lab) had responsibility for software running on MVS, including IMS. After evaluating the possibility of integrating relational capability from System R into IMS, they decided to keep them separate and develop plans to build a new RDBMS on all major mainframe platforms (MVS, VM, and VSE). However, their budget and skills would limit them to building only the MVS version.

Around the same time, the IBM lab in Endicott, New York, which developed software for smaller mainframes running VM and VSE, had just canceled a project to merge the VM and VSE operating systems and had more budget and skilled programmers available than they had projects. So, the decision was made to focus the development of the VM and VSE RDBMS...