What are SQL Server Reporting Services Shared Datasets?
In SSRS, a "shared dataset" is exactly what it sounds like: a query that can be used by more than one report. Much like views in SQL Server, shared datasets can be helpful in two main ways:
Encapsulating query logic: in a higher education setting, for instance, let's say we have four different reports (all in the same SSRS project) that are used to track information specific to a current initiative--in this case encouraging diversity on campus. Further, let's say that, for the purpose of this initiative, we are using some special student demographic information that we don't typically use in our reporting. With SSRS shared datasets functionality, we can create a single query that returns the student demographic fields that we are interested in and then "share" that query among all four of our reports.
Performance optimization: again, like a SQL view, a shared dataset can be tweaked for performance. We can establish a special caching plan for long-running queries, for example.
Potential Gotchas
One thing to remember when creating shared datasets is to set the SSRS project-level variable "OverwriteDatasets" to true, because you will typically want the ability to update query logic in your datasets and then deploy those changes to the SSRS server.
WSU-Specific Notes
To see SSRS shared datasets in action, open up the Summer Housing project--all of the reports here use at least two shared datasets.
More Information
http://msdn.microsoft.com/en-us/library/ee633675(v=sql.110).aspx