I've used it out of curiosity and it works, but honestly I just fell back to my primary weapon of choice: Pentaho. This isn't because I'm not comfortable searching data at the command line, it's more because of momentum in that all my other data is already in a database and I usually need to join datasets, not just search them.

It's trivial to create a two hop transformation in Pentaho. The first step reads your data out of a CSV, the second loads it to a table. Pentaho will generate the DDL needed to create the table based on discovered field names and types found in the import step based on data introspection.

Having said that I've often wished for a tool exactly like this that's lighter weight than drill (and doesn't expose my data on a web page like drill does) so I will definitely bookmark this for the next time it's needed. Hopefully it makes it into the Redhat EPEL.