Curt Monash

The Gnarly History of IBM's DBMS Product Line

By Curt Monash
May 20, 2005 12:22 AM EDT
My immediately prior blog entry ascribes a certain bogosity to IBM's embedded DBMS market share numbers, because of the diversity of cats and dogs in its product line. One might argue that this isn't IBM's fault, and one would have a point -- were it not for the fuss IBM makes about overall DBMS market share numbers. They even more or less confessed at the time of the Informix acquisition that a strong motivation was a simple desire to build reported market share. So, from memory, here's a surely incomplete history of IBM's DBMS product line. Long long ago, in a computing environment far away, IBM offered a forerunner to today's MRP/ERP systems called BOMP, for Bill of Materials Planning. BOMP begat D-BOMP, an abstracted version that has some claim to being the first commercial DBMS. From this grew IMS and DL/1; DL/1 later faded away, but IMS remains as an important legacy "hierarchical" DBMS. Meanwhile, E. F. "Ted" Codd begat System R, which begat (actually, grandsired) DB2, which begat a separate version of DB2 on UNIX boxes. Meanwhile, there were a bunch of relational DBMS startups, generally focusing on VAXen or UNIX boxes as their initial platforms. These included Relational Software, which became Oracle; Relational Technology, which became Ingres; Relational Database Systems, which became Informix; Data Language, which became Progress; and Sybase, which became Sybase. (And be grateful that I'm sparing you a lot of other name changes.) By the mid-1990s, Oracle was the industry leader; Ingres had fallen by the wayside; Sybase was in the process of imploding; Progress was focused on the low/midrange; Microsoft was just starting out; and IBM couldn't get its act together. Informix, however, was doing quite well at the high end, and seemed to be emerging as a clear #2 to Oracle, based on its clean, simple, powerful DBMS architecture. Hah. A buying spree ensued. Informix was already working to combine two major code lines -- Informix Classic, and the high-end XPS line picked up in Beaverton, OR when Ingres suicidally spun it off. That in itself wasn't all bad -- Oracle itself was strengthened by the acquisition of DEC's Rdb technology -- but they didn't stop there. First of all, they bought Illustra, and now were integrating three code lines into their core product. They also managed to pick up Red Brick for data warehousing (although XPS was a fine product for that already), another data warehousing operation, Cloudscape for mobile/Java DBMS (a real nonstarter), and so on. They even bought what itself was a cobbled together company that amounted to Ascential-plus-Universe, where Universe was more or less the old Pick operating system and its built-in DBMS (I must confess to being fairly unclear on the details of that part). When Informix's management was removed for a combination of fraud and general stupidity, what later became the Ascential management team briefly ran the whole now-bicoastal company. Soon, however, they sold the DBMS to IBM, spinning off Ascential as a middleware company with a billion or dollars in the bank, that IBM just wound up buying later. And that, children, is how IBM came to be a DBMS industry market-share leader.