As sad as it is, there needs to be focus on one way forward - that appears to be tabular. A migration path is needed, but having the two models is confusing and becoming unnecessary. Even if full support for MD features is not available I think this should still happen.
5 Comments
- pedro8New MemberThis is an extremely obtuse assertion. Multidimensional cubes are a different architecture which has its place in the enterprise and more often I find myself having to deploy both tabular and multidimensional models. Tabular has some advantages but it haves some serious limitations as well, and this is simply the way it is. In fact the more I use SSAS in tabular mode the more I find its limitations to be infuriating. One might as well say they want to deprecate pencils because of pens. If MSFT ever deprecated MD models they would be losing their competitive advantage to the likes of IBM and Oracle.
- Adrian_TheumaNew MemberI agree Pedro. I can't see this ever happening without Tabular first catching up to MD.
- fimianiNew MemberAgreed. There are a number of gaps that MD has that Tablular does not, and a few in the other direction as well. Tabular shortcomings when compared to MD: * write-back, * non-leaf writeback, * many-to-many, * unary operators, * several other critical things... Tabular upside compared to MD: * ability to on-the-fly model (add user-defined measures, dimension, granularities, hierarchies), * faster query times so I've heard, * other stuff? By no means an all-inclusive list. All that MS has to do is build out a gateway on top of MD that serves up the MD data as flattened Tabular slices, and there wouldn't be any issue here. Guess I'm just compelled to use Tableau until they get all of this resolved.
- pedro8New Member'@tf: Another serious limitation for me is the Direct query functionality of tabular models, which is meant to be analogous to ROLAP cubes. The fact is that when you turn direct query on, we get a severe set of limitations, which in my view makes Direct Query as useful as an ashtray on a motorbike :) For example we cannot use direct query when combining multiple data sources. This means that creating a tabular model to query line of business applications is inpractical since I cannot combine it with other data-sources, such as a date dimension from another DB.
- fbcideas_migusrNew MemberStatus added:New
Recent ideas
Sparklines: customizable data point limit (allow more than 52 points)
Sparklines in table and matrix visuals currently display up to 52 data points per sparkline. We'd like to suggest a customizable data point limit, so report authors can allow more than 52 points and ...v-kkalvala4 hours agoMicrosoft EmployeeNew14Views2likes0CommentsAllow dynamic headers for Calculation Groups
calculation Groups reduce duplicated DAX, but their Calculation Item names become static Matrix headers such as Current, MTD, QTD, and YTD. These headers do not tell report consumers which measure i...danieleloFabric7 hours agoNew MemberNew9Views1like0CommentsCentralized retention policies for imported data in Power BI semantic models
Summary Please provide a tenant-level retention policy that allows administrators to centrally enforce automatic deletion of expired data from Power BI Import semantic models, independently of data ...JINHOYOO7 hours agoMicrosoft EmployeeNew64Views19likes1CommentResizable Personalize visual pane or wrapped field names
Problem: The Personalize Visual pane width is fixed and long measure/column names are cropped making it difficult for users to identify which item to add to their visual. Idea: Allo...Simon_AUS13 hours agoRegular VisitorNew46Views8likes0Comments