SQL Server Data Tools (SSDT) 16.3 was made available last week and features a new Analysis Services Tabular Model Explorer for SSAS 2016 (1200 compatibility level) models. Details, screenshots, and guidance for future enhancements to this feature were posted by Kay Unkroth at the SSAS Team Blog: Introducing Tabular Model Explorer.
There are already many good reasons to upgrade Tabular models to 1200 compatibility level including Bi-Directional Cross Filtering, Display Folders, Calculated Tables, Translations, Row Level Security and Calculated Columns for Direct Query models, and a much more fluid, richer design experience inside of SSDT. The addition of Tabular Model Explorer makes the upgrade even more compelling and complements all the new features of SSAS Tabular.
I’ve spent only a few hours with the Tabular Model Explorer so far and as someone whose spent many hours working with large Tabular models I’m very, very pleased by this new feature. Since released for SQL Server 2012 Tabular Model projects inside of SSDT have been essentially very similar to Power Pivot for Excel (plus partitions and row-level security roles). This makes perfect sense when migrating a Power Pivot model (and hopefully soon a Power BI Desktop model) to SSAS Tabular and is generally sufficient for small to medium sized projects (e.g. 1-2 fact tables, 100 or less metrics). However, for larger, more complex models that SSAS Tabular 2016 is better positioned to support, analyzing and implementing changes can be a tedious and time consuming task.
Top/Favorite 16.3 Tabular Model Explorer Features:
- At the highest level I just like having the Tabular model objects centralized and organized in a tree structure. Rather than clicking throughout model and bouncing between the different windows (given the dependencies between model objects) the model author/designer can more easily remain at an abstract level and quickly analyze different objects and properties. The feature will also be of significant value to SSAS developers that are new to Tabular models and not familiar with navigating to the different windows and dialogs.
- The Search Box will be very useful with large models and I like how this feature aligns with the search box in Power BI Desktop
- I also love how clicking an object such as a column in the Tabular Model Explorer immediately highlights and/or selects this object in the data window – this one feature alone makes analyzing a Tabular model much easier and faster.
- The Right-click context menu for the Project provides very quick access to common functionality. (I’d like a Deploy option here like in Solution Explorer window but it’s a very nice menu nonetheless)
Future or Potential enhancements I’d like:
- Display folder structures used for metrics. With 1200 models you can now assign metrics to display folder structures but you can’t currently see the metric folders as they would appear to the users without leaving SSDT/VS and viewing the model from Excel or Power BI Desktop.
- A parent folder for Partitions. Just like Measures have their own parent folder I think, again in the context of large Tabular models, easy access to a model’s partitions (rather than drilling down through the tables) would be a nice addition.
- Common tasks that currently require a model author to leave the Tabular Model Explorer interface could be added via right-click or other option. For example currently you can’t create a new hierarchy from the Model Explorer and you can’t make a change to two columns (ctrl+shift) such as hiding two columns like you can from the data window.
- A right-click option on the Measures parent folder that brings up a DAX measure authoring window (with folder structure, Intellisense, function list, etc) would be great as well – developing DAX metrics inside the data grid still feels a bit too much like Power Pivot.
- An ‘Analyze in Power BI Desktop’ option would be nice too as this will increasingly be the common client tool.
Again, overall this is a fantastic new feature and I’m excited to see future Tabular and DAX related enhancements in SSDT and SSMS.