| 講演抄録/キーワード |
| 講演名 |
2008-09-21 13:00
[招待講演]Advanced Query Optimization Techniques in a Parallel Computing Environment ○Wook-Shin Han(Kyungpook National Univ.) DE2008-30 |
| 抄録 |
(和) |
We introduce two query optimization techniques that we recently developed, which can be employed in a parallel computing environment. The first technique is called "progressive optimization in a parallel database"; the second one "parallelizing query optimization." Progressive optimization (POP) is a technique to detect cardinality estimation errors by monitoring actual cardinalities at runtime and to recover by triggering re-optimization with the actual cardinalities measured. In order to apply POP to a parallel environment, 1) rigorous voting schemes are used to make a global decision of whether re-optimization should be triggered; 2) a mechanism that re-uses a distributed temporary table is used. To reduce optimization time in multi-core processor architectures, a novel framework for parallelizing query optimization is introduced. This framework permits for the first time exploiting dynamic programming to find optimal plans for some complex OLAP ad-hoc queries referencing more than about 12 tables. |
| (英) |
We introduce two query optimization techniques that we recently developed, which can be employed in a parallel computing environment. The first technique is called "progressive optimization in a parallel database"; the second one "parallelizing query optimization." Progressive optimization (POP) is a technique to detect cardinality estimation errors by monitoring actual cardinalities at runtime and to recover by triggering re-optimization with the actual cardinalities measured. In order to apply POP to a parallel environment, 1) rigorous voting schemes are used to make a global decision of whether re-optimization should be triggered; 2) a mechanism that re-uses a distributed temporary table is used. To reduce optimization time in multi-core processor architectures, a novel framework for parallelizing query optimization is introduced. This framework permits for the first time exploiting dynamic programming to find optimal plans for some complex OLAP ad-hoc queries referencing more than about 12 tables. |
| キーワード |
(和) |
query optimization / progressive optimization / parallel databases / / / / / |
| (英) |
query optimization / progressive optimization / parallel databases / / / / / |
| 文献情報 |
信学技報, vol. 108, no. 211, DE2008-30, pp. 1-2, 2008年9月. |
| 資料番号 |
DE2008-30 |
| 発行日 |
2008-09-14 (DE) |
| ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
| PDFダウンロード |
DE2008-30 |