Home // International Journal On Advances in Software, volume 10, numbers 3 and 4, 2017 // View article

Measurement-based Cost Estimation Method for Multi-Table Join Operation in an In-Memory Database

Authors:
Tsuyoshi Tanaka
Hiroshi Ishikawa

Keywords: Non-volatile memory; In-memory database systems; Query optimization; Query execution cost

Abstract:
Non-volatile memory is applied not only to storage subsystems but also to the main memory to improve performance and increase capacity. In the near future, some in-memory database systems will use a non-volatile main memory as a durable medium instead of the existing storage devices, such as hard disk drives or solid-state drives. For such in-memory database systems, the cost of memory access instead of I/O processing decreases, and the CPU cost increases relative to the most suitable access path selected for a database query. Therefore, a high-precision cost calculation method for query execution is required. In particular, when the database system cannot select the proper join method, the query execution time increases. Accordingly, a database join operation cost model using statistical information measured by a performance monitor embedded in the CPU is proposed and the accuracy of estimating the change point of join methods is evaluated. The results show that the proposed method can improve the accuracy of cost calculations to more than 90% compared to the conventional method. In conclusion, the in-memory database system using the proposed cost calculation method can select the best join method.

Pages: 459 to 476

Copyright: Copyright (c) to authors, 2017. Used with permission.

Publication date: December 31, 2017

Published in: journal

ISSN: 1942-2628