/
lectern.tv
Play15s30s1x0:00/--:--
DoneWatched
Expand
lectern.tv
18:45/53:57
1:19
2:13
0:56
0:38
0:51
1:45
0:30
0:41
5:29
1:05
3:26
6:39
3:26
2:44
1:41
2:09
3:23
2:05
2:03
1:50
6:24
1:33
SQL Optimization Background
Discover why query optimization is the unsolved art and science that separates breakthrough database engines from the rest.
This introductory lecture outlines the query optimization course at Carnegie Mellon University, detailing the fundamental challenge of generating efficient execution plans for declarative SQL queries. Andy Pavlo covers the DBMS component pipeline from parser to physical executor, explains the difference between heuristic and cost-based optimizers, and compares top-down versus bottom-up optimization strategies.
Referenced inCMU Database Group · CMU 15-445 (Fall 2022)
Lecture Guide
- 1Declarative languages like SQL require a query optimizer to automatically find a correct and efficient execution plan.
- 2Small changes in query plan decisions, such as join order or operator choice, can yield performance differences of multiple orders of magnitude.
- 3The core database pipeline consists of a parser, binder, system catalog, optimizer with a cost model, and physical executor.
- 4Query optimizers generally rely on either heuristic rule-based rewriting or cost-based search algorithms.
Proofs, Examples & Demos
Assumed Background