Partitioning and Bucketing in Hive

In this recipe, we will look at Partitioning and Bucketing concepts in Hive. Partitioning is a scheme by which we reduce the amount of a dataset to be queried for a result.

For example, let's say we keep a record of the entire population in a Hive table, and when someone asks for a person's details, we query it based on the name and gender. This will take a lot of time, as it has to go though many records. Instead, a better approach would be to partition it by country and then query only the specific subset of records. In the RDBMS world, this is known as sharding. Partitioning lets you organize the data in a logical fashion and skip all but the data you need.

Bucketing is a concept of hashing, and based, upon the principle of hashing, it places the data into various buckets, which are themselves files. In simple terms, it is clustering the columns in a group set so as to reduce the number of partitions. Having a large number of partitions is also a bottleneck...