Gunther is right, Hive planner rewrites distinct using group by, so it doesn't matter what do you use from performace point of view. ... View more

@Ravi teja You can get this information from the metastore database: hive=> SELECT "TBL_NAME", "COLUMN_NAME", "TYPE_NAME" FROM "TBLS" , "COLUMNS_V2" WHERE "TBL_ID"="CD_ID" UNION SELECT "TBL_NAME", "PKEY_NAME", "PKEY_TYPE" FROM "PARTITION_KEYS" p , "TBLS" t WHERE p."TBL_ID"=t."TBL_ID" ORDER BY "TBL_NAME"; TBL_NAME | COLUMN_NAME | TYPE_NAME ------------+-------------+----------- tableA | aaa | string tableA | bbb | string tableB | foo | string tableB | bar | int tableC | cola | string tableD | colb | string (6 rows) ... View more