Pushing Query Processing Back to Source Systems

During the scheduled refresh of semantic models retrieving from on-premises sources, any query transformations not executed by the source system will require local resources of the M (Mashup) engine of the on-premises data gateway server. With larger semantic models, and potentially with other scheduled refreshes occurring on the same gateway server at the same time, it becomes important to design M queries that take full advantage of source system resources via query folding. Although transformations against some sources such as files will always require local resources, in many scenarios, M queries can be modified to help the engine generate an equivalent SQL or other source system statement and thus minimize local resource consumption.

In this recipe, a process and list of items are provided to identify queries not currently folding and the potential causes. Additionally, a query based on an existing SQL statement is redesigned...