Power BI Shared Datasets vs Cubes
Published
March 31, 2020
Microsoft continues to develop their business intelligence stack, recently taking the next step to create a concise and unified ecosystem by enabling shared datasets. For businesses that rely solely on the Microsoft stack for their warehouses, SQL Service Analysis Services (SSAS) and Visual Studio have been the rule of thumb for data modelers, but Microsoft is changing this paradigm.
Microsoft has released permissions that allow Power BI datasets to be published out to the organization, groups or individual users. Data models can also add tags such as “Promoted” and “Certified” to move these shared datasets to the top of the list and allow the organization to provide some differentiation between the sets as they see fit.
This new feature moves warehouse development a step closer towards one ecosystem. Now modeling and report creation can now be done and managed within one application. For example, if a new column is added to a dimension table in SSAS and needs to be added to a report then the designer needs to (1) connect to the cube in Visual Studio, (2) refresh the data source or table, (3) add new column to model, (4) save or publish the cube, (5) close Visual studio, (6) open the power bi desktop file, (7) refresh connection to cube, (8) edit report and (9) publish report. With Shared Datasets, modelers can simply open the power bi file containing the dataset and refresh your dataset. If that file is the same one that contains the report, then you are all ready to make your edits. No need to jump through hoops and switch applications. Simplification is now the name of the game.
Another benefit of shared datasets is the progress towards achieving a “single source of truth.” Datasets can now be shared across workspaces so that different departments or types of users get their information from the same source. Visual tags let the user know which dataset should be used. This will help to curtail the use of rogue sources and unsubstantiated numbers within an organization, while also providing users the freedom to build new reports knowing that the data has been vetted and validated.
A final minor benefit of shared datasets is enhanced management of the gateways. The gateways are the tool that pulls from the data source and refreshes the dataset. Previously, a different connection was needed for each cube. Now, all you need is a connection to the data warehouse, providing less overhead and only one connection to manage and maintain.
When a brand-new piece of functionality replaces a long-standing one, it is important to be aware of potential trade-offs. The long history of SSAS has led to development of a lengthy list of features at the developer’s hands. One key element missing from Shared Datasets is the robustness of its security. SSAS allowed for the creation of row, column and even cell-level security that could be tweaked by a modeler. With shared datasets, row level security can be handled with roles but compared to the flexible and dynamic nature of SSAS, it can seem lacking.
Another consideration in the jump to Shared Datasets is the performance. Time will tell if models created in Power BI will be able to have the same throughput, calculations, and performance seen by SSAS. Finally, as with any new tool, it is important to consider the support it will receive. Whether it is widely adopted, and Microsoft continues to add to it, or it is abandoned, and the model creation will need to be overhauled again has yet to be seen.
We’re excited to see how shared datasets enables better data uniformity and shared ways of working, and happy to help your organization explore this new feature.
OneSix is a Power BI Consultant
We help companies solve their most complex problems with cloud technology and yield incredible results.