Tuning of the barrel points HIVE

• Points barrels rules

  • Of sub-barrel field values ​​hash, hash value divided by the number of barrels remainder, the remainder of which determines which records in the bucket, which is the same as the remainder in a bucket

• Points advantage barrel

  1. Improve efficiency join query:

     Tables A and B is assumed for join, join the id field conditions:

     • Two tables for the big table

     • Two tables are divided bucket list

     • A number of buckets is a multiple of the number of table B or table tub factor

     So join query time, each bucket of Table A and Table B can directly join the corresponding barrels, instead of the entire table join, to improve query efficiency

  2. Increase sampling efficiency

     • Points table behind the barrel can not bring on the field name, the default when the band is not by sub-barrel field, could bring, but no sub-barrel table you must bring

     • When sampling by sub-barrel field, because the partition table is to go directly to the corresponding bucket of the bucket to get the data in Table relatively large increase sampling efficiency