ML Wiki

Query Processing Pipeline

Steps:

1. Translating SQL to Relational Algebra
2. Logical Query Plan Optimization
   • finding cheaper equivalent expression:
   • translating to Conjunctive Query and simplifying it
   • applying some heuristics (pushing selections and projections, etc)
3. Physical Query Plan Optimization
   • doing the cost-based assignment of physical operators
   • assigning each node of Logical Query Plan a physical operator
   • for that need to know statistics from Database System Catalog
4. execution engine - for that important things are:
   • Physical Data Organization (databases)
   • Indexing (databases)

Sources

• Database Systems Architecture (ULB)