A lot of that is true, but to be sure there are use cases for NoSQL--they are simply too often misunderstood or abused, being used when a traditional RDBMS woul

A lot of that is true, but to be sure there are use cases for NoSQL--they are simply too often misunderstood or abused, being used when a traditional RDBMS would have worked fine. Elastic/horizontal scaling for certain types of data that are amenable to it (i.e. event logs, feeds, append only type stuff): its going to be hard to keep up with something like Cassandra there, where you can pretty much add servers on the fly to pick up load. In memory stuff for sure, for a distributed volatile in-memory key store. Simple flexibility you get going schema-less: if you have an embedded DB/on-premise application, you can't just get into your customer's environment and perform a migration; having a flexible schema can make patch updates easier.