O'Reilly logo
  • Karthik Ramachandran thinks this is interesting:

The simplicity of a dimensional model also has performance benefits. Database optimizers process these simple schemas with fewer joins more efficiently. A database engine can make strong assumptions about first constraining the heavily indexed dimension tables, and then attacking the fact table all at once with the Cartesian product of the dimension table keys satisfying the user's constraints. Amazingly, using this approach, the optimizer can evaluate arbitrary n-way joins to a fact table in a single pass through the fact table's index.


Is there a point to be made here about how this pattern maps to MapReduce