ssas | Server side application skeleton for Go language | Server Side Rendering library

In MDX, WHERE conditions are normally done on attributes, not on measures. So, the best approach would be to make an attribute from your Row Count column (i. e. making it part of a dimension).

Then, you could either leave your calculation as it is, enabling users to do their own where conditions, i. e. by dragging Row Count either into a filter or to rows o columns to see the breakdown by row count.

If you want to filter in your calculation itself, then to add the row count filter to the calculation, you would use tuples wherever you use just a measure currently, e. g. (assuming the attribute is called [Dim1].[Row Count] in the cube, and hence the member for row count 1 would be [Dim1].[Row Count].&[1]), you would use

Try with this:

First, I think that is a bad IDEA to use CALENDARAUTO because it searching for all date columns from your model (and if you have a Customer with a born date eg. 1912-02-02, then you create a big table from that date).

CALENDARAUTO ignores calculated tables and calculated columns searching for date columns. Only the imported columns are analyzed to search for date columns.

Internally, CALENDARAUTO calls CALENDAR providing a date range that include all the days in the range of years referenced by data in the model, according to the following rules:

The earliest date in the model which is not in a calculated column or calculated table is taken as the MinDate. The latest date in the model which is not in a calculated column or calculated table is taken as the MaxDate. The date range returned is dates between the beginning of the fiscal year associated with MinDate and the end of the fiscal year associated with MaxDate.

A better idea is to use CALENDAR where you have more control

Power BI's strength is data visualisation, and it is likely to be well suited for for using on top of you retail data warehouse.

I'm not sure which research paper you are referring to, but Microsoft has been topping Gartner's Magic Quadrant for Analytics and Business Intelligence Platform for several years now, followed by Tableau and Qlik. If you are interested in reading further around the various platforms, you can download from https://info.microsoft.com/ww-Landing-2021-Gartner-MQ-for-Analytics-and-Business-Intelligence-Power-BI.html?LCID=EN-US

Power BI does not require SSAS or SSRS to run. If you already have SSAS, Power BI can use SSAS as a data source, and it works very well with a live connection, alternatively you can model the semantic layer directly within Power BI itself. Power BI, especially now Paginated reports are included is seen as a cloud based alternative to SQL Server Reporting Server

Yes you can use time intelligence calculations in SSAS. The best place to start for checking DAX support in the version is Power BI, Excel and SSAS is dax.guide, here for example is the DAX link for TOTALYTD

You can create a Date Dim in a number of ways in a DAX format for example using CALENDARAUTO, or import it from you calendar in your source system if you have one. However as you are using SSAS compatibility 1200, you don't get the option of calculation groups (Compatibility 1500), which means you'll have to create a YTD measure for each of the items you need a YTD value for. For more on this subject please see the SQLBI resource on time intelligence

It turns out that in our case it was probably caused by cache being dropped each time the cube was processed.

Our solution would be creating SSIS Package that would run certain DMV queries to populate cache every time we process our cubes, so end-users would be able to use the cache instead of generating it themself.

Automatic processing might have some issues when SSAS tries automatically identify objects that need to be processed and what type of processing is needed.

It seems that the dimension table with dates is been handled but the related fact table is not. If you have the option, you could set full processing as separate steps. For example, do full processing to dimensions on the first step and then full processing to fact tables on another step. Lastly do calculate (aka. recalc) to the whole cube. That will save some amount of memory. If not enough though, you could set different steps for each fact table. Remember to do calculate at the end.

At this moment SSAS Tabular & Azure Analysis Services doesn't support the Aggregation Feature, that is in Power BI Desktop and the service

The underlying engine for SSAS Tabular, can be split into two functions, the storage engine and the formula engine. When queried, it basically gets the data from the storage engine, and passes it to the formula engine to run the DAX queries. You are right, in that the formula engine is single threaded, but the storage engine is multi threaded. So the storage engine will quickly get the data, then the formula engine will do the calculations on it. But multiple users queries, will result in multiple single threaded queries, and not the same thread being used, so it will in affect parallel process the queries.

If you look at the spec of the recommended size of an instance is around 8 to 16 cores, as most queries will not be running at the same time and the storage engine will be running as well as the calculation one at the same time, SSAS will quickly return data and only take seconds to process. The most import part will be the RAM limits that you have to store, and run queries.

If you look at Power BI Premium you'll see 4 cores, with 25GB of ram, which can take 1200-1500 report interactions per hour. With Azure Analysis Services get a measure of QPU's which equals about 1 virtual core to 25 QPU's. For S2 that is the same a 4 virtual cores.

For reference you should check out the following blog posts on the vertipaq engine and the SQLBI post here, by Marco Russo, one of the leaders in SSAS Tabular and DAX, on the considerations of sizing your server.