Oracle Loader for Hadoop can be used to load data from multiple data sources (for example Hive, HBase), and data in multiple formats (for example Apache weblogs, JSON files). There are two ways to do this:
These approaches are discussed in our Oracle Open World 2013 presentation.
(1) Use an input format implementation. Oracle Loader for Hadoop includes several input format implementations. In addition, a user can develop their own input format implementation for proprietary data sources and formats.
(2) Leverage the capabilities of Hive, and use Oracle Loader for Hadoop to load from Hive.