CA Releases Ingres r3 Database as Open Source

by Timothy Prickett Morgan

Software wants to be free, some people say, and beleaguered software giant Computer Associates, which has been embroiled in an accounting scandal for more than a year, was able to get some good press this week as it took its Ingres relational database, which is popular on Unix servers, out to the open source community as an open source product for Linux, Unix, and Windows platforms. While the Ingres software will be free, CA is hoping to make some money on different support options.

Ingres is not new to the open source community. The Ingres database is older than dirt, which means it is roughly concurrent with the Unix operating system. In fact, Ingres was one of the early relational databases, coming out of a research project at the University of California at Berkeley (just as did the Berkeley Systems Design, or BSD, variant of Unix that is at the heart of FreeBSD, NetBSD, OpenBSD , and Sun Microsystems' Solaris, which is also about to go open source this month). The open source version of Ingres was widely known as University Ingres or Berkeley Ingres, while ASK Computer Group commercialized Ingres. As Unix took off solidly in the mid-1990s, CA paid $310 million to acquire ASK (in 1994, to be precise) and therefore get its hands on a Unix database, which it needed to sell alongside its IDMS mainframe database. Incidentally, ASK was not the only company to commercialize code from the Ingres project, which was released under the BSD license. Sybase (and therefore Microsoft's SQL Server), Informix, Tandem's NonStop SQL (for fault tolerant machines) are all derivatives of Ingres. (The people who worked on the original Ingres project founded Sybase, Ingres Corporation (later eaten by ASK), and Tandem, in fact.) And the kicker Berkeley project to Ingres, jokingly called PostgreSQL, is today a popular open source relational database, too. It is clear that the open-sourcing of Ingres has already created a huge wave of innovation for the past 25 years.

By taking Ingres open source--or, more precisely, taking the latest CA version of Ingres back to the open source community--CA is trying to do two things. Ingres was never a household name and is arguably more obscure than Oracle, IBM's DB2, SQL Server, the open source MySQL, and Sybase. CA is betting that by allowing customers to adopt Ingress as an open source component of their IT infrastructure and by allowing them to contribute to the future development of its product, it can offload some of the Ingres development burden while at the same time fostering an installed base of loyal users. Having done that, a certain percentage of those loyal customers will want to buy support services for the open source Ingres they roll it into production. This is exactly how Linux, JBoss, MySQL, and other open source components have been commercialized even as they are wickedly commoditized. You give away the software, and extract support fees for the minority of the base that needs it.

CA is releasing Ingres r3, as its open source variant of Ingres is called, it under its own license, which it calls the CA Trusted Open Source License. This license, which is a derivative of the Common Public License, allows software developers to embed Ingres r3 into their products and to distribute it freely so long as they make the Ingres source code available as part of their solution. CA has pumped $1 million into a developer's contest and will award prizes to developers who create cool solutions that help customers of other databases port their code to Ingres r3. In effect, CA is paying the open source community to create porting tools and relying on the desire for cash as well as hubris to motivate developers. This is a wise move.

Ingres r3 is available immediately on Linux and Windows, and you can get download it here. Early next year, Ingres r3 will be available on Solaris, HP-UX, AIX, Tru64 Unix, and OpenVMS. The Ingres r3 database includes database clustering for high availability and horizontal scaling built in, and has table partitioning and indexing for big databases commonly used for data warehousing. The software also includes parallel SQL query processing, which significantly speeds up ad hoc database queries, and online table and index reorganization. The database will run in 32-bit or 64-bit modes on all platforms.