Scaling the storage typically implies a lot of architecture decisions to stay away from transactional constrains of relational databases. In this brave new world of could services, these decisions are not optional anymore. The good news is that, once we start developing for the new storage engines, we may get contaminated and end up wearing a nosql t-shirt.
In this session, José António Silva showed why tables, queues, key-values and other patterns will help us use scalable distributed persistence stores like Microsoft’s Azure Storage and Azure SQL.
An older webcast was recorded by Microsoft and can be found here http://canoas.com/blog/2009/05/18/Webcast+On+Cloud+Database+Options.aspx