We recently upgraded our cluster to 6.0.1 / MEP 5.0.0 and are trying to modernize storage of our primary datasets. One of these is Netflow data, which is stored in a MapR-DB JSON table. (Bad choice??) I'm able to do very crude searches using mapr dbshell, but previously used Impala. All indications are that the cool kids are all using Drill; but having crashed and burned attempting this 2 years ago (See: Drill can't see MapR-DB tables - which was never resolved, despite being marked as 'answered') and reading and re-reading the docs, I'm not optimistic I can make this work in my lifetime.
So should we fall back to Impala - or are there other options with better performance that we should consider?
Thanks for any input from anyone actively using MapR-DB JSON tables.