Dr. Michael Stonebraker, the innovator behind Ingres and Progress databases, argues in his blog, The Database Column, that massive data engines the likes of SQL Server and Oracle are a model of “One Size Fits All – A Concept Whose Time Has Come and Gone.” His new start-up, Vertica, is offering a “column store” database engine, in contrast to the older model “row store,” and the idea is appealing: why store a heterogenous group of data items together because they logically make up a “record” of data, useful to the users, but of no convenience to store together for the computer, when instead you can store a column of values that spans some spectrum of values, and create links to the records to which they apply? I’m simplifying the concept, I’m sure, but Dr. Stonebraker offers some interesting performance claims on how his engine can address big-iron problems faster and more efficiently.
Keep an eye on this technology.