Summary

An MDC table is just like any other table; it can be partitioned with a partitioning key, and that partitioning key may or may not also be one of the table's dimensions. An MDC table can have views, MQTs, referential integrity, triggers, RID indexes, replication, etc., defined upon it.

In an environment with referential integrity relationships, either or both of the parent and child tables may benefit from block-level clustering, depending on the query workload against these tables and the nature of the data contained in them. Similarly, in a star-schema environment, the fact table may benefit from block-level clustering, as may one or more of the dimension tables. Each table should be evaluated separately, and the dimensions should be chosen based on the query workload, data distribution, and expected cell density.

With the correct choice of dimensions for the MDC tables, the unique and powerful performance enhancements that multi-dimensional clustering provides can be leveraged.