set hive.new.job.grouping.set.cardinality=2; set hive.map.aggr=true; set hive.groupby.skewindata=true; CREATE TABLE T1(a STRING, b STRING, c STRING) ROW FORMAT DELIMITED FIELDS TERMINATED BY ' ' STORED AS TEXTFILE; -- Since 4 grouping sets would be generated for the query below, an additional MR job should be created -- This is not allowed with map-side aggregation and skew SELECT a, b, count(1) from T1 group by a, b with cube;