I want to get your feedback on storage models we are planning to use in SSAS cube design. For the current year or last one year of data, we are planning to use MOLAP. For the previous years, I am considering HOLAP, as we expect queries on aggregration levels of attributes, and do not want to support any drill through and browsing below the aggregation level.

The reason I am asking, in the storage video, I see that HOLAP is not recommended.

The other question, once the storage model is MOLAP and HOLAP as defined above, I could remove the relational DB access from SSAS, correct ?