Working under the application programming we saw, that ADO.NET Entity Framework was a data source. We also saw that, at platform level, the class containing business operations was allocated. Such allocation in a separate class allows us to abstract certain client technologies and data sources. So, data sources can be datastores based on POCO, services and datastores with custom logic. Similar reasoning is fair for a client part. The .NET appendix (Silverlight, WPF, etc), JavaScript code, services, etc. can be a client for such service. In fact, building business logic once means it can be used repeatedly.

Figure 5: Overview of concept of .NET RIA Services

Besides that, allocation of a central object with business logic that does not depend on representation and data source makes it possible to build unit tests for it and to develop such applications in Test-Driven-Development style.