Single-direction cross-filtering relationships are the most common in Power BI datasets and particularly for data models with more than one fact table. In this dataset, whether import or DirectQuery, all relationships are defined with single direction cross-filtering except for the relationships from Sales Territory, Product, and Date to their corresponding bridge tables, as described in the following section on bidirectional relationships. 

The following screenshot from the Relationships View includes three of the seven dimension tables related to Reseller Sales:

As you can see from the arrow icons in the Relationships View, the filter context in single-direction relationships exclusively navigates from the one side of a relationship to the many side. In the absence of any DAX expressions that override the default cross-filtering behavior, tables on the one side of single-direction relationships are not filtered or impacted by filters applied to the table on the many side of the relationship. For example, the Employee table has 299 unique rows based on its Employee Key column. A measure, such as Count of Employees, that references this column will always return the 299 value despite any filters applied to other tables in the model. 

There are, of course, valid business scenarios for allowing the filter context of the related fact table to impact dimension measures, such as the Count of Employees or the Distinct Count of Product Subcategories. Dataset designers can support these requirements by default via bidirectional cross-filtering relationships, but in most scenarios this isn't necessary or appropriate. For these DAX measures, the CROSSFILTER() function can be applied to override the default single-direction cross-filtering. See the following CROSSFILTER section for the function syntax and a use case example.