Take for example an external table called “blog” created with the following partition scheme:ĬREATE TABLE blog (title STRING, body STRING, pubdate DATE) PARTITIONED BY (dy STRING, dm STRING, dd STRING, dh STRING) In these cases, the partition location may not conform to the default dynamic Hive partition location (which takes the table's base location and appends a name=value path token for each partition), and can take any valid data path as the location for the partition.
However, partitioning is also useful for external tables where the data may already reside in HDFS or be managed by a service other than Hive. Partitioning Hive tables is a great strategy to improve query performance for Hive-managed tables by avoiding full-table scans. With the latest Hue release, the metastore offers better controls over partitioned Hive tables.
