The feature, developed by Google, is nicknamed Shards. It provides a way to map data across multiple instances of a database that share common schema.
This approach to scaling a database is often called horizontal clustering. Shards allows you to map data across horizontally clustered databases that may be portioned, so that one instance covers North American records and another is the counterpart for EMEA.
For instance, if you need to do a lookup on customers from databases, with separate instances for North America and EMEA, Shards lets you view them as a single virtual database against which to query or to insert data. It also allows you to conduct parallel queries with functions that you used to pay for a la carte with the specialized parallel database editions that providers like Oracle and Informix used to charge top dollar for.
JBoss has also added support of unstructured text searching, and is bundling an API to the Apache Lucene open source test search engine. However, if you want to use another text search engine, it will support plugging in other APIs.
Finally the new version also adds more integration between different Hibernate components, reducing legwork that developers must perform when installing the framework. And JBoss has improved compatibility with IBM WebSphere, BEA WebLogic, and other pure Java Persistance API (JPA) tools.
