Just finished reading the latest issue of IEEE Annuls of Computing History, volume 31 number 4, on the history of database systems. It covers the early history of the CODASYL-style databases like IMS, IDMS, Adabas, and Total in the 1960’s & 1970’s.
As a database professional who started with relational products Ingres and Oracle back in the early 1990’s, my whole view of databases has been RDBMS centric. It is easy for me to think that Oracle, DB2, and SQLserver dominate the database world and all databases are SQL.
It has been just fascinating to read about what came before the RDBMS. These are the products that established the concept of a database as separate from the OS and in many ways also established the idea of a software product as separate from software that was just included “free” with the hardware. I say they are CODASYL-style because none of them eventually implemented the full CODASYL spec though IDMS is closest as it was used as a basis for the spec. IMS is heirarchical so it was farthest away but still the same style of DML by direct record access rather than querying sets like SQL.
As an RDBMS DBA, it is a bit humbling to realize that these so-called “obsolete” products like IBM IMS and CA-IDMS still process most of the world’s transactions today. They move all this data without SQL of course, these databases are older than SQL, something the new “NoSQL” crowd should keep in mind. They are hardly obsolete either, IBM for example, just released a new version 11 of IMS, with even Java and SOA support.