You need to slice data by the CouponUsed, OnSale, and Closeout columns. What should you do?

A linked dimension is a dimension that is shared across multiple measure groups or cubes; it does not address the problem of storing flag columns from a fact table.

A degenerate dimension holds a key attribute (such as an order number) directly in the fact table with no associated dimension table, which is not applicable to multi-column flag attributes.

A role-playing dimension is a single physical dimension used multiple times in a fact table under different aliases (e.g., Order Date and Ship Date both pointing to a Date dimension), which does not apply to unrelated flag columns.

CouponUsed, OnSale, and Closeout are classic low-cardinality boolean or flag attributes with no natural parent dimension. A junk dimension combines all permutations of these flags into one small dimension table with a surrogate key, replacing three separate foreign keys in the fact table with one and keeping the dimensional model clean.