As companies look to service-oriented architecture as a way to effectively coordinate services with partners, suppliers and customers, the importance of ensuring data accuracy and consistency is magnified.

The vision of an enterprise data fabric is to provide real-time data services that keep everything used in an SOA transactionally consistent while performing the necessary data transformations between systems. The data fabric is complementary to an enterprise service bus (ESB) and, in effect, will enable data integration just as the ESB enables process integration.

The data fabric virtually connects not only the back-end relational data stores that feed systems within an enterprise but also the distributed, high-performance, in-memory data caches that are embedded within systems such as Web services. Heterogeneous support is a critical aspect of the data fabric, because most organizations not only have multiple database technologies in place but also rely on a number of application platforms, including mainframes.

The software infrastructure used to create the data fabric has design time and run-time components. The design time provides a GUI environment to define mappings between schemas of the different data sources and create a Common Information Model (CIM). Bidirectional transforms would then be specified, tested and debugged before being deployed into the run-time environment.