A second possible approach would be for the middle tier to query a view created inside SQL Server. The view should be responsible for communicating with all the different data sources, merging the results and providing a single result set to the calling application.
As shown in Figure 8-2, by moving the responsibility of merging the results to SQL Server, we can take advantage of the following facts:
The middle tier doesn’t have to be coupled to the physical implementation.
SQL Server can provide a single result set, regardless of its physical distribution, so the middle-tier code is much easier to write and less risky in terms of security, maintenance, concurrency control, and communication.
By processing ...