I'm new to the MapR ecosystems - we are looking to migrate our current BI platform from Oracle to Map-R and we want to give it a try using our transaction data. To add more context, we first need to import all the data for the past 10 years but eventually we intend to stream data into MapR using MapR-ES (or Kafka) if we do stick with MapR. Looking at the products Map-R has to offer we are trying to figure out:
Should we import data directly into MapR-fs using the NFS mount or should we store the data into MapR-DB as json documents.
Secondary indexes in MapR-DB look appealing and will allow for random (based on known fields) searches by the fraud team however most of the direction from the docs is that MapR-DB is geared more for CRUD apps vs MapR-FS which is geared more towards batch jobs.