Models & patterns of non-relational databases
Many companies are starting to follow the large Web players (Google, Yahoo, FaceBook, Amazon &c) in adopting scalable non-relational databases, that can handle the extreme demands of web-scale. New "nosql" databases not only address these demands, but also do it efficiently without the huge traditional costs of RDBMS clusters & storage solutions.
Here's a nice explanation of the different models & patterns of these databases:
For more info, check-out the slides & videos from the recent nosql-east conference (the speakers are the authors of most of these databases):