Way back in 1999 I tried to shed relational database by using quite a few good object relational databases. It was a liberating experience, and I grew and learned by making some bad mistakes, including making the data unsearchable. I think that is a lesson I can carry forward when looking at the new breed of high availability, on-demand scalable candidates. You do have to build your own indexes for text search.

Today we have more reason to look beyond relational databases. They just don’t work that well in the cloud.

At SynthaSite, we will be building prototypes around some of the candidates out there. These include Mnesia, SimpleDB, CouchDB as well as abstractions like Elastra. We will also be taking a long hard look at using data store alternatives if the requirements are simple enough. MogileFS is a candidate in this regard.

I am working with a great team, which helps when grappling with these things. In particular I am enjoying working with Neil for whom I have had great respect over the years, but until now I have not had the opportunity to work closely with.

Neil is also the one who gave me shove to start blogging again.