Hi, Current Carbondata has deep integration with Spark to provide optimizations in performance and also supports features like compaction, IUD, data maps and metadata management etc. This type of integration forces user to use CarbonSession instance to use carbon even for read and write operations.
So I am proposing standard spark's FileFormat implementation in carbon for simple integration with Spark. Please check the jira for the design document. https://issues.apache.org/jira/browse/CARBONDATA-2872 -- Thanks & Regards, Ravindra
