For a partitioned Hive table (stored as ORC), I can count the rows in a partition very quickly with a query like this, presumably because Hive gets the count directly from table statistics:
select count(*) from db.table where partition_date = '12-01-2015'
How can I just as quickly get counts from multiple partitions? A query like this launches a full tez job and takes a couple dozen seconds to run depending on the date range I choose:
select partition_date, count(*) from db.table where partition_date >= '11-01-2015' group by partition_date
Thanks!
I am running Hive 0.14 if that is relevant.