I still hear people say that data virtualization is just a new name for data federation--old wine in new skins. Not true! If we look at the current generation of data virtualization servers, we have to conclude that they offer a lot more functionality than data federation servers.

Data federation is run-time technology that makes it easy for an application to access a heterogeneous set of data stores. In this case, the data federator deals with all the different API's, the different database languages, it will try to optimize access to those data stores by doing distributed join optimization, and it will handle all the issues of distributed transactions.

In my book on data virtualization (Data Virtualization for Business Intelligence Systems), I define data federation as follows:

Data federation is an aspect of data virtualization where the data stored in a heterogeneous set of autonomous data stores is made accessible to data consumers as one integrated data store by using on-demand data integration.

Data virtualization is much more than data federation. Here are some of the features supported by data virtualization servers today:

• Self-service, iterative, and collaborative development
• (Canonical) data modeling
• On-demand data profiling and data cleansing
• Full support for the entire development life cycle: business glossary, information modeling
• Extensive data integrity features
• Extensive master data management features
• Integration of different data integration styles, including ETL, ELT, and replication

In a nutshell, where data federation is primarily run-time technology, data virtualization supports the entire system development life cycle. So, it supports modeling and design as well. Or, to you use popular terminology, data virtualization = data federation++.

If you want to know more about this topic, attend my session at the The Data Virtualization Experts Forum.