Cluster is very easy in MongoDB and for unstructured data. I've looked into clustering postgresql a few time and it seems like a pain in the butt. TBH, storing data without thinking about relation is just kicking the can down the road.

My previous company store coupon code on it (referrals). I prefer a hash nosql for that though (cassandra) but for a one man shop it's overkill.

---

edit:

IIRC postgresql have their own document storage like MongoDB. I just never had to use such thing because I guess I'm too OCD about how my data suppose to look like.

Relational DBs for everything that needs joins. MongoDB is horrible with that and should not be used. One would not put an off-road vehicle on a race track and expect it to win.

I use MongoDB for everything where I don't know the data structure before hand. MongoDB is very good in indexing unknown data structures and makes them searchable. Also, when I build prototypes I often use MongoDB.

The biggest one is searching jsons. You essentially have a database that stores jsons and you can search it. Just dump a json and search it goes a long way. My problems are more the atomicity guarantees which are not very consistent.