Abstract: The dimension is a key component of data warehouse, and it is the primary query object in OLAP. However, the standard star/snowflake schema has distinct limitations for modeling the dimension hierarchy in practical applications, and SQL doesn’t provide the direct support for dimension and its hierarchy, which makes the query expression of OLAP verbose and unreadable. In this paper, the SQL（D） data model based on the relational database is presented, which gives the formal definition of hierarchy tree, hierarchy chain and dimension. The unbalance and heterogeneity of dimension hierarchy are permitted in this model. The SQL is also extended efficiently to support the definition of dimension, the comparison of dimension’s levels, the reference of dimension, and the appointment of dimension level to roll up, thus significantly simplifying the query expression of OLAP. Finally, the implementation algorithm of the extended semantic is given.