Some honestly aren't much more... they all have at least a "key" or id field you can lookup a record by, and most have a map/reduce implementation. Some require you to do your own indexed records and others (like MongoDB and RethinkDB) have the ability to add additional indexes (similar to SQL).
I find that MongoDB tends to perform at least as well as an SQL based solution in general use. Some use cases are better suited to SQL (anything with multiple records in a single transaction as a hard requirement for example). It really depends on your needs.
I find that MongoDB tends to perform at least as well as an SQL based solution in general use. Some use cases are better suited to SQL (anything with multiple records in a single transaction as a hard requirement for example). It really depends on your needs.