Data Virtualization has several capabilities that allow it to perform different roles in delivering information agility:
Therefore it is often not possible to fix its place in the enterprise architecture. Rather it varies depending on a broad spectrum of use case patterns it supports (see patterns and architecture for Analytical, Operational, Web/Cloud, Data Management use cases).
However the "hour glass" architecture below provides a broad generalization across use cases. Data Virtualization plays a role in the bottom, middle and top of the enterprise information architecture and it can do all three simultaneously:
Image based on Hourglass architecture in the Forrester´s "Data Virtualization Reaches Critical Mass" report
Data virtualization is essential for business agility. But it does not replace the need to persist historical data, connect applications, improve source data quality, and so on. It has to work with other data management and middleware tools to bring out the most cost-effective and agile solutions to problems. Here are some ways it does that: