Simulation/Spark

Educational simulation
Spark

Catalyst Optimizer

Unresolved → analyzed → optimized → physical plan.

Watch the canvas:logical stagephysical planLive simulation
1. Unresolved logicalnames only, no types2. Analyzed logicalcatalog resolved, types bound3. Optimized logicalpushdown, pruning, folding4. Physical planFileScan → Exchange → HashAggregate
Catalyst is four trees. Each one is a strictly better description of the same query.