Metric Views is not a direct cube engine. It's a semantic abstraction layer. It allows to define dimensions, fact, measures with aggregations and then query it flexibly across tools.

You get multiple choices

• Star Schema Approach

  Adopt a star schema (Fact & Dimension along with joins in metric views) design as it simplifies metric view creation with easier governance and extensibility

• Modular Metric Views- Create separate modular metric views based on purpose (Sales by State - Metric View with a join between sales and state dimension, Sales by Maker - Metric view with a join between sales and product dimension). It aligns well with domain driven design and easy to manage

More details here