Home // DBKDA 2025, The Seventeenth International Conference on Advances in Databases, Knowledge, and Data Applications // View article


On the Performance of Query Optimization Without Cost Functions and Very Simple Cardinality Estimation

Authors:
Daniel Flachs
Guido Moerkotte

Keywords: query optimization; hash join; cardinality estimation; cost function; plan generation.

Abstract:
In order to enable a fast time to market for new Database Management Systems (DBMS), we introduce two simple, very easy to implement cardinality estimators and a build-plan method that does not require any cost function. Experimentally, we demonstrate that different plan generators incorporating these ideas are quite competitive on the Join Order Benchmark (JOB): the join ordering algorithm DPccp yields plans that are at most a factor of 2.10 away from the optimum without using any runtime-dependent cost function if cardinalities are known. Thus, using our approach obviates the effort of implementing sophisticated cardinality estimation methods and cost functions in a first version of a DBMS.

Pages: 58 to 64

Copyright: Copyright (c) IARIA, 2025

Publication date: March 9, 2025

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-68558-244-9

Location: Lisbon, Portugal

Dates: from March 9, 2025 to March 13, 2025