Share your thoughts on the new model view UI (preview)

[tessahurr]

Hit Reply and let us know what you think of the new model view UI.  To learn more about this feature, please visit our blog.

Here are some areas that we'd like to hear about in particular:

• Performance and the experience for large models that exceed 50 tables
  • Edit: We've heard some feedback regarding the All tables tab in the model view. If you are using a model with 50+ tables, you can mitigate some of the issues you may be facing by setting the environment variable PBI_diagramTableThreshold to a number greater than or equal to 10 and less than 1000. This environment variable allows you to adjust the number of tables that the All tables tab will show.
• Table card tooltips and the information they display when you hover over a table header

Thanks and we look forward to hearing your feedback!

- The Power BI Model View Team

[educ8tom]

Where to set the PBI_diagramTableThreshold environment variable?

[RWright]

I really don't like it. I can't find what I am looking for. Buttons are placed over ellipses within a visual so you can't use the drop down. It's like learning Power BI all over again.

Much harder to use.

[marcritchie]

Every time I make changes in Transform data, e.g. adding a new Query, the changes aren't reflected in Model view but are shown in Report view and Data view

I'm using the latest version of PBI Desktop (2.95.804.0 64-bit (July 2021)), I have less than 10 tables in it. I am at the point that I have to close the file down and open it again every time that I make any changes to the underlying data which is very time consuming.

[pade]

I would like to see a Collapse all button someware. This would make it easy to create a relationship diagram without having to show all attributes. Yes this will also require a Expand all button as well.

And at the same time having a "Hide hidden columns" and a "Show all/hidden colums" button would also be nice 😉

[Pat_msft]

@alec20191104 , @cacfab , @Anonymous 

In the december release that went just out, the default limit has been increased to 75. We are working on further improving the experience here. 

[Anonymous]

Trash my table relationships view without warning and with no recourse??  Absurd and ridiculous.  For performance reasons?  Heck, just delete my data model and all the data.  Zero results with zero latency.  Perfect performance.  Bonuses all 'round.

And I get to spend my time searching on how to correct this absurdity and then more time trying to rebuild the relationships in a year-old data model with quite a few tables.  All because I dared to add a new table using the "latest" PBI Desktop release.

Can you tell that I'm not pleased with this "enhancement."  "Auto Detect" was useless.  I have to rebuild all the table relationships manually? Really.  Whoever came up with this as an "improvement" should be ashamed.

[Anonymous]

So the Evironment Variable setting appears to work an restore functionality.  I would recommend in the future "enhancements" that you make the "old" system the default, and have any "new" features like this require effort on the part of the user to enable the feature whenever the "new" feature disables existing functionality.  I'm back online sooner than I expected, but the disruption was a disruption of flow and productivity nonetheless.

[brian_haney]

Thank you for the opportunity to provide feedback on the new model view UI. I have been a Business Intelligence consultant for a few (15) years, so hopefully my input will add value.

For a given table in the schema view:

Add formatting to make the visual quicker to interpret:

• Key Fields and Join Columns (Bold Font - Sort to top - Use Key like icon)
• Hidden Fields (Italic Font)
• All other fields (Normal Font)
• Add Column Data Type (Date, Integer, Text, Real, etc.) and Format (Currency, Decimal, Percentage, etc.) to each field (helps me know when I join a date stored as a date to a date stored as an integer).
• In addition to a hidden icon, need a null (blank) icon. It is important to know which fields have blank values so I can coalesce it and not have to constantly answer questions from the customer about "what does (blank) mean?" Argh!
• Add Number of Records (In Import Mode) after Table Name (e.g. Shipments (Records: 678,999)) or at the bottom of the table. This shows if you're working with full or partial data, and if I have Table A with 4 million rows and Table B with 6 million rows, it is a quick sign NOT to join the two.

Ability to sort Folders by ordinal position (Customer is more significant than a Bin).

Ability to sort Fields by ordinal position not alphabetical (Start Date displays before End Date).

Keep the Property -> Advanced Group open when selecting other fields.

In the Property for field, show the fully qualified source field (Ex. Target: Shipment Number <-- Source: erpdb.oper.shpnbr)

Make Reference Tables (i.e. Role Based Dimensions) obvious as well as include the table that it references. Maybe a light grey background or grey font.

Ability to create a relationship using a "BETWEEN" relationship. For example I have a Inventory Stock Level of 100 units starting on 1/10 until 1/14. On 1/15 I receive a shipment of 50 units. I cannot create a relationship that states Calendar Date BETWEEN Inventory Level Start Date and Inventory End Date. Gotta do that, but supporting complicated DAX or M-Query adds an unnecessary level of complexity.

The ability to create filters as fields. For example if I only want Active Customers, I can create a Filter object that drags to the filters not visual to limit results. So if I only want to see Active Customer, I can create a filter field that does this. This is valueable for creating complex DAX objects out of the box so that end users don't have to learn DAX.

The ability to export the model view metadata to CSV, JSON, or Excel. 

• Folder Name, Ordinal Value, Description, Flags, Storage Mode, etc.
• Field Name, Ordinal Value, Descriptions, Flags, Formats, Data Category, etc.
• Join / Relationships: From Table.Column, To Table.Column. Join Type (Equi-join, Between, Fuzzy match, etc.)

Predefined Metadata Fields

• Data Source Name
• Data Last Refresh Date
• Number of Rows
• Next Refresh Date/Time (Incremental Loads)
• Parameter Fields & Values

Add Perspectives. Selecting a perspective highlights the tables in that perspective. Selecting multiple perspectives highlights all the tables in those perspectives.

Create a Lineage View. We know that the organization in the source is technology driven, while the organization in the semantic model is business driven. There is not a 1:1 match from table -> folder and column -> field. A lineage view would show the transformations used to get from the source to the target. (Steal it from Power Query -> Query Dependencies or the dataflow view). This should include Aggregations in the lineage.

Leverage the empty space in the bottom status bar to tell me about my model:

• File size
• Last saved date/time
• Compression rate
• Data sources
• Last refresh date/time
• etc.

[Pat_msft]

@Ole111 - What did you set the value to? it must be set to an integer between 10 and 1000. I would very strongly discourage setting it higher than 100 though. If you have more tables in your model than that number, the All Tables layout will be disabled.

[Ole111]

@Pat_msft - it was 100 (as you suggested as an example).

[Anonymous]

I have the same gripes as others. Removing the All Tables view without an easy button "Enable All Tables" and having the Model Diagram re-appear just as we have had it, meticulously ordered and structured so that we have a clear diagrammatic view of everything... I have to say... 

FYI creating a new layout, CTRL ALT dragging everything onto the canvas and re-ordering everything just doesn't cut it. No longer can we have new tables automatically appear on the canvas, which means there is that overhead of periodically having to check that all tables are actually on the canvas, especially when you open up a report that another person has developed....

[barbforsman]

How/where do you set the environment variable? 

[pjserlet]

Enter "System variables" in the search box in Windows.

[pjserlet]

Hi @tessahurr 

If I select my table in the Model view, I can't switch the storage mode from DQ to dual. The option is not visibile in the properties of the table.

Is there a solution for this ?

Regards

PJ

[pjserlet]

This was resolved for me by rebooting my computer... 🤔 

[Anonymous]

Please please bring the All tables View back. 

You may give us option to build new layout with less tables, but don't disable the schema already built.

Please please roll back!

[Anonymous]

This is what I got by ctrl A dragging.

[Anonymous]

When I click reset layout, it looks better. I am ok now.

Thanks,

[Anonymous]

From there I started to remove small tables.  Having a all table view is comforting.

[Anonymous]

This new feature is absolutely useless and disappointing - I didn't notice any improvement in the performance. Moreover, it shouldn't be turned on by default - basing on the experience of other commenters I believe there could be serious issues with the data model and existed relationships because of this "improvement".