You already said you will be looking and SQL closer, so let me give a few pointers here. Consider that in SQL you specif how to store data, including indexes,

You already said you will be looking and SQL closer, so let me give a few pointers here. Consider that in SQL you specif how to store data, including indexes, and then what you wan to get, but not how to get it. In other words you specify data structures and the engine picks a set of algorithms and builds an execution plan. The next logical step would be for the user to specify the schema, while the database engine is not only choosing algorithms but also reshapes your data into data structures that would allow optimal algorithm selection later on. At first it would only build new indexes, but then it could also start employing partitioning and ultimately move to normalization/denormalization. Secondly, be on the looklout for when a problem reminds you of a SQL query. For example, a mail application looks like a query where you specify which fields from which emails you want to see and in what order, and on top of that query result you're specifying a style sheet, CSS on steroids, to make it render pretty. Style sheet is also declarative, not imperative! I am presently working on a framework that follows this model to facilitate iPhone app development, and it's looking pretty good already.