Simpli-Squared: Optimizing Without Cardinality Estimates

Most query optimizers rely on cardinality estimates to optimize their execution plans. Traditional databases such as PostgreSQL, Oracle, and Db2 utilize synopses, such as histograms, samples, and sketches. Recent main-memory databases like DuckDB and Heavy.AI often operate with minimal or even without estimates, yet their performance does not necessarily suffer. To the best of our knowledge, no analytical comparison has been conducted between optimizers with and without cardinality estimates. In this paper, we present a comprehensive analysis of optimizers that use cardinality estimates and those that do not. To represent optimizers that don't use cardinality estimates, we design a simple graph-based optimizer that only utilizes join types and table sizes. Our evaluation on the Join Order Benchmark reveals that cardinality estimates have a marginal impact in non-indexed settings, whereas inaccuracies in estimates can be detrimental in indexed settings. Furthermore, the impact of cardinality estimates is negligible in highly parallel main-memory databases.