Deprecate multi dimensional model

[simon54]

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.

[pedro8]

This 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_Theuma]

I agree Pedro. I can't see this ever happening without Tabular first catching up to MD.

[fimiani]

Agreed. 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.

[pedro8]

'@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.