An evaluation of PERF joins for a two-way semijoin based algorithm.
Date of Award
CC BY-NC-ND 4.0
Distributed database system is becoming more widely used instead of centralized database systems in business world due to business expansion and network technology development. Query optimization provides a strategy for executing each query over the networks in the most cost-effective way, which aims to minimize the transmission cost over the networks. Many techniques and algorithms have been proposed to optimize queries, such as semijoin[BC81][BGW+81], 2-way semijoin[KR87], composite semijoin[PC90], hash semijoin[TC92], PERF join[LR95], etc. In distributed query processing, the semijoin has been used as an effective operator to reduce the total amount of data transmission. 2-way semijoin is an extended version of semijoin for more cost-effective distributed query processing. PERF joins are 2-way semijoins using a bit vector during the backward phase. PERF[LR95] is designed to minimize the cost of the "backward" reduction. It is based on the tuple scan order instead of hashing. Thus it does not suffer any loss of join information incurred by hash collisions. Algorithm UPSJ and Algorithm CPSJ are proposed based on a 2-way semijoin algorithm. Two variants of PERF joins are applied to the 2-way semijoin algorithm. In Algorithm UPSJ, uncompressed PERF joins and 2-way semijoin techniques are combined. In Algorithm CPSJ, compressed PERF joins are applied during the backward processing. Programs are designed to implement both original and the enhanced algorithms. Several experiments are conducted and the results showed a considerable enhancement obtained by applying the PERF join concept.Dept. of Computer Science. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2005 .Y36. Source: Masters Abstracts International, Volume: 44-03, page: 1419. Thesis (M.Sc.)--University of Windsor (Canada), 2005.
Yang, Li., "An evaluation of PERF joins for a two-way semijoin based algorithm." (2005). Electronic Theses and Dissertations. 1010.