I've had lots of fun using Mongo with Spark, I have found it pleasant, though most Spark libraries are good to use, including the JDBC ones. I haven't had much luck with the kind of viz where you don't first do ETL from Mongo to some tabular format.