Hive is more adaptable as far as data arranges that it can check - You may see Hive as more component wealthy as far as SQL language support and inherent capacities - Hive will probably finish your inquiry regardless of whether there are hub disappointments (this makes it reasonable for long-running employments); this is valid for both Hive on MR and Hive on Spark - If Impala can run your ETL, at that point it will most likely be quicker - Impala will come up short/prematurely end a question if a hub goes down during inquiry execution - The last point may make Impala less reasonable for long-running occupations, obviously there is likewise a shorter disappointment window since questions are quicker, so Impala might just suit your ETL needs on the off chance that you can endure the faiure conduct ... View more