The Teiid project (http://www.jboss.org/teiid/) is a data virtualization system allowing applications to use data from multiple, heterogeneous data stores. These data sources currently include a broad range of relational databases, web services, text files, and ldap. Every data integration project has a unique set of requirements often with multiple way to approach solving the problem. Some use cases are best addressed with a classic data warehousing approach by transforming and integrating data (ETL) into a centralize DW for reporting and analysis. In other cases, a query federation approach is more applicable where the source data remains where it is, and a query federation engine like Teiid is used to present multiple, disparate datasources as a single entity. In fact, we believe there are all use cases that can benefit from combination of these technologies.
Teiid provides the an optimized engine for unifying data sources without moving data.
Pentaho Data Integration provides an easy-to-use, robust library of transformation objects that can be used to transform data so that its easier to integrate with other heterogeneous data sources in Teiid.
As a DBA, I want the ability to incorporate data from xyz source with my data warehouse without having to make drastic changes to my data warehouse.
As a Virtual Database modeler (Teiid user), I want a robust transformation library so that I can make appropriate modifications to my data source before incorporating it into my Virtual Database.