3.3 Enable Access to Hybrid Partitioned Tables

Using hybrid partitioned tables (HPT) enables efficient and economical information lifecycle management. Data copied to HDFS and Oracle Object Storage can be referenced from HPT. Archive data is saved to HDFS - providing scalable, inexpensive storage.

For example, you may have a table partitioned by month. "Hot" data that is queried frequently and subject to updates would be stored in partitions that are managed by Oracle Database. The older historical data can then be off-loaded to Hadoop.

Query performance benefits from data that is partitioned. When querying data for a particular month, you only need to query a single partition - not the entire table containing data for every month. Queries against recent "hot" data will not even attempt to access the archive data in Hadoop; the archived partitions are pruned at query time.

The data in Hadoop can be stored in various formats - including parquet, CSV, etc. - allowing you to easily use that data by Hadoop tools and applicatons.

For a complete description of how to use HTP, see Hybrid Partitioned Tables in Oracle Database VLDB and Partitioning Guide.