One of the most astounding bit of news to come out of this week's Google earnings call is the fact that the firm had to leave $1bn worth of advertizing on the table for lack of available inventory. It shows that as fast as Google is growing, the market is growing even faster! This is why Google has been trying hard to pull as much content as possible onto the Web, be it public content by making it easier for us to publish with tools like Blogger, Google Pages and Google Base or private data with applications such as Gmail, the just released Google Calendar and the recently acquired Writely.

But the most promising thing to happen this week was the publishing of the GData API and it's first derivative, the Google Calendar API. This is really the first step in building the ultimate Atom Store, the mother of all databases. When the relational model was invented 30 years ago by Ed Codd at IBM a lot of people were deriding the approach as too simplistic and not powerful enough. But the beauty of the approach was the realization that in order to scale you had to simplify the whole framework: power through simplification.

A similar approach is now being taken by proponents of the Atom Publishing Protocol which GData is building upon. The relational approach which allowed Oracle to build a powerhouse doesn't scale very well beyond the enterprise. Adam Bosworth, the father of Microsoft Access and now VP of engineering at Google has been advocating a complete redesign of current databases for a while now and according to Dare Obasanjo GData has Adam written all over it.

If the web was able to achieve its phenomenal growth it is because it is mostly read only. Scaling of the web was achieved by using RESTful protocols and extensive caching, something that current databases are quite bad at. In order to achieve the same scale for a writable web, a new approach needs to be taken and Google is betting on the Atom Store. This is bold. As bold as moving from Network and Hierarchical models to the Relational model. That could spell the death toll for databases as we know them and that could prove to be even more painful for Microsoft et al. than the competition on the Office front. This is the O/S attack that many have been predicting but it's not the me-too product that some have been describing. And this one could actually be even more devastating.