set hive.map.aggr=true;
set hive.groupby.skewindata=false;
set mapred.reduce.tasks=31;

CREATE TABLE dest1(key INT, value DOUBLE) STORED AS TEXTFILE;

EXPLAIN
FROM src INSERT OVERWRITE TABLE dest1 SELECT src.key, sum(substr(src.value,5)) GROUP BY src.key;

FROM src INSERT OVERWRITE TABLE dest1 SELECT src.key, sum(substr(src.value,5)) GROUP BY src.key;

SELECT dest1.* FROM dest1;