How the World Will Change Databases

One of the most fascinating lectures Ive attended lately was “How Databases Changed the World,” put on by the Computer History Museum. The event last week featured a panel discussion by the biggies in the database business, including Chris Date, Michael Stonebraker (formerly of Illustra and Informix, among others), Ken Jacobs (Oracle), Bob Epstein (formerly of Sybase) and Pat Sellinger (IBM).

Its beside the point that no one really discussed how databases changed the world. They discussed how the database world changed when it moved from the Cullinet hierarchical system to the relational systems in use today.

The relational model was one of the most substantial computing changes ever to occur. It not only led to the development of analytics, data mining and packaged applications, but also to the separation of data from those applications, which, in turn, created the fundamental model of how we compute today.

It took 15 years from the time E.F. Codd and Date described the relational model before databases based on the technology became mainstream. Since that time, everyone has been looking for new models. Object databases, for example, were thought to be the next shift in computing in the early 1990s. XML data stores are now thought to be included in the next wave of database technology.

Unfortunately, database technology has hit a plateau. The best vendors can do is increase speeds and attempt to extend the relational model to include user-definable extensions that will allow it to prevail for the foreseeable future. Technologies for parallelism and clustering are necessary and technically sophisticated, but theyre lacking in vision.

The big question is if were ready for the new wave of applications. First, in the medical field, miniature cameras are taking megabyte-size pictures every few seconds in millions of patients. Radio frequency identification tags are beginning to include time, geospatial data and more in real time. Billions of digital images are taken every year. But where are these images and data stored? They are in the file system, not the database.

The database is supposed to be able to handle, compare and become the platform for analytical applications for this data. But its not going to happen any time soon—the relational model is stuck.

When will the databases grow up? Write to me at john_taschek@ ziffdavis.com.