I can only partially answer your question, but hopefully more than a comment.
MongoDB is usually not a repository for keys and, as you know, has certain characteristics when used as one.
MongoDb also has a lock issue that might come back to haunt you. It has a DB level lock, which means it can (require testing) cause write lock saturation.
It is also heavily designed to read 80% of the application (which is said to be the most common installation for websites at the moment), so the more you write, the more you will notice performance degradation over time. This suggests that you can configure MongoDB to be more write-friendly, and distributed nature helps to slightly stop recording saturation.
However, to say that my personal opinion is the SQL MongoDB learning curve:
- Close to
null - More natural and easier to implement in my application than SQL
- The query language is simple to make it dead easy to access
- The query language has much in common with SQL
- The drivers are standardized so that the syntax that you see in the Documents for the JS driver on the console is consistent across all sections.
My personal opinion on a general issue is its common concept. If you get a NoSQL solution designed for key stores, then that can be really good. A quick Google search pulled out a small list of NoSQL key value stores on Wikipedia: http://en.wikipedia.org/wiki/NoSQL#Key-value_stores_on_solid_state_or_rotating_disk
source share