Extending Role Security in Analysis Services for SQL Server
By using and implementing a many-to-many relationship in a cube, existing tables with user account information can be used directly to secure cube contents, rather than converting to role based security in the cube. You can also secure your analysis services more easily and access sensitive information contained within it.
The goal of this paper is to describe a method that allows users to browse an Analysis Services cube, limiting the user's access by the user account used to connect to the cube. The method makes use of a many-to-many relationship established in the underlying online transaction processing (OLTP) database. Instead of maintaining many roles and assigning users to roles, the technique described uses a table in the data warehouse to link users or groups to the parts of a dimension to which they are allowed access. In a large organization, this simplifies security maintenance greatly.
Given an OLTP schema, implement a data warehouse schema with tables that connect a user account using a many-to-many relationship from the user account to a dimension in the table. A cube is then created from the data warehouse which contains the many-to-many relationship. A role is created to implement dimensional security to limit what members of the dimension the user is able to view.
Diagram 1 shows the data warehouse schema and the relationship between the Security table and the Product dimension. This database was created by extracting appropriate data from the Northwind database, which is one of the sample databases supplied with SQL Server in the past. Northwind is a small, simple database that is useful for testing and prototyping. The Northwind database is still available from CodePlex as a downloadable script. After creating a simple data warehouse with a single-fact table and three dimensions (time, customer, and product)-two tables were added to implement the security model. Windows user accounts are listed in the Security table and linked to the Product dimension through the ProductSecurity table.
The contents of my security table have two user accounts, Bill and William. The ProductSecurity table will link the user accounts to the dimension surrogate key, DimProductKey, in ProductDim. Bill is mapped to all products except to Beverages, whereas William is mapped to just Beverages. The simplicity employed here will make it easier to verify the results when I apply the role in the Cube browser.
The data source view brings in all the tables from my simple data warehouse. Because of the many-to-many relationship between the ProductDim and Security tables, I will build the first part of the project using the Cube wizard and the three-dimensional tables (CustomerDim, ProductDim, and TimeDim).
The Basic cube will need a measure group that links the ProductDim and Security tables. The relationship between ProductDim and Security is a many-to-many relationship in the underlying data source view, so a measure group is created that is mapped to the ProductSecurity linking table.
Next, we will add the security dimension.
We now have a cube structure with a measure group mapped to the SalesFact table, a measure group mapped to the ProductSecurity table, and four dimensions (Customer Dim, Product Dim, Time Dim, and Security).
Once the Product Security measure group and Security dimension are part of the cube structure, we must define the dimension usage in the Dimension Usage editor, which is the second tab from the left in the cube editor.
Both measure groups are in place. A relationship must be added to connect the measure groups in a many-to-many relationship, which ties the user ID in the security table to the product key in Product Dim. The Product Security measure group serves as the linking element in the many-to-many relationship. Click on the ellipsis button to bring up the relationship editor.
The interface is simple. Choose "Many-to-Many" as the relationship type and choose "Product Security" as the intermediate measure group. This will then link the user ID in the security table to the Sales Fact table through the Product Security measure group. The Product Security measure group is in place simply to allow this relationship to be defined. The cube should process and deploy at this time. You can browse the cube in the cube editor browser. You should implement some dimensional hierarchies; this step is left up to the reader.
Create a role. I renamed mine as Master.Role. In the Cube tab of the Role Editor, set the access attribute to Read. Then move to the Dimension Data table. In that tab, choose Security from the Dimension combo box selector, and enter the MDX statement in the Allowed member set text box, as shown in Diagram 11. Check the box at the bottom of the interface, Enable Visual Tools.