Follow @DBDebunk Follow @ThePostWest

Until the rise of nosql solutions, you would force-fit all data and relations into traditional RDBMS model and then do a fire dance in the application layer to manage them. In other words, you would force a cube into a cylinder with enough force that it fits. With nosql solutions, that is not necessary, so you can pick the right solution based on the problem domain. If your data is largely schemaless, there is no point in forcing it into a relational table with most of the columns being null, so a document oriented DB such as MongoDB or CouchDB would be ideal. If you dealing with lists, sets, queues etc., which need to be persisted, then Redis maybe an ideal candidate. If you are managing key/values then Cassandra maybe your choice. Of course, you don't want to have a dozen of these in your system, but a careful use of RDBMS and nosql solutions would greatly simplify application design.
--Have any NoSQL databases become viable replacements for SQL databases yet?, quora.com