What MDX can and SQL Can't perform on the same datatbase [closed]

It's not a matter of what one can do vs the other, as you can always, in theory, write SQL to replace your MDX. It's a matter of how hard it is.

Example:

If you want the sales per month of your top 5 clients from last year in EMEA, your MDX query is something like

SELECT TopCount( [Clients].Members, 5, ( [Measures].[Sales], [Time].[2018] ) ) on Rows, [Time].[2018].Children on Columns from [Sales] where [Geography].[EMEA] 

and your cube definition will take care of what columns to query from which tables, getting the Territory name from the dim_territory table, the customer name from the customer table, etc.

The equivalent in SQL would be something like

Select c.customer_name, t.month_name, sum(f.sales) from fact_sales f, dim_customer c, dim_time t, dim_geography g where f.time_id = t.time_id and f.customer_id = c.customer_id and f.geography_id = g.geography_id and g.territory = 'EMEA' and t.year= 2018 and c.customer_name in ( select c.customer_name, sum(f.sales) from fact_sales f, dim_customer c, dim_time t where f.customer_id = c.customer_id and f.time_id = t.time_id and t.year = 2018 group by t.year limit 5 ) group by t.year, t.month_name order by t.month_number 

Not only the SQL query is much more complex to write (I'm not even sure what I wrote is correct, tbh), but a lot of variations that in MDX just require a small rewrite, in SQL they will require several nested queries, and lots of filtering, grouping etc.

Plus, MDX allows you a cross-tab result, where you get one column for each month (or whatever dimension level you choose), whereas in SQL it will return a normalised dataset and you'll have to pivot it on the client side if you want it in tabular format.