So it depends... And my list isn't all inclusive/exclusive, just a general starting point I like to use. I generally use SQL when I have really well defined r

So it depends... And my list isn't all inclusive/exclusive, just a general starting point I like to use. I generally use SQL when I have really well defined relationships that change rarely and where I have data that needs to be shared amongst many disparate systems (especially if I don't control one or more of them). NoSQL I use more as an application specific data source and cache. Many times this means warming up the NoSQL from the relational side for some data elements. I really don't like to pollute a SQL datastore with application specific components unless they need to be shared and I don't control all the applications. I don't agree with people that draw the line at the need for transactions. Transactions can be managed in either SQL or NoSQL and even across them both with some care.