[daveman]

As an analytics professional who was pressured into a MongoDB environment, I feel the OP's pain. If you want to do gymnastics with your data, (aggregations of aggregations, joining result sets back onto data), SQL expressions are a 1000 times easier than Mongo constructs (e.g. map reduces). We usually ended up scraping out data from Mongo and dumping records into a SQL database before doing our transformations.

All that said, our developers loved the ease of simple retrieval and insertion, and of course the scalability. So I guess you ultimately need to base your decisions on your priorities.

I don't fault the OP though, since it's hard to know just how limiting NoSQL will be until you try to do all the things you used to assume were database tablestakes (no pun intended).