Dynamic Load Balancing Method for Parallel Database Engine on Heterogeneous Computing Resources
-
- OKUNO Akihiro
- The University of Tokyo
-
- HAYAMIZU Yuto
- The University of Tokyo
-
- GODA Kazuo
- The University of Tokyo
-
- KITSUREGAWA Masaru
- The University of Tokyo,National Institute of Informatics
Bibliographic Information
- Other Title
-
- 共有ストレージ型並列データベースエンジンにおける演算資源非均質性の影響を軽減可能とする動的負荷分散手法
Abstract
In parallel database engines running on heterogeneous computing resources, if works are assigned to nodes equally, processing time on nodes with lower capacity limits overall query execution time, and consequently, the capacity of rest nodes cannot be utilized fully. The public cloud environment, which has been widely used, provides virtualized computing resources to users via network. However, it is somtimes impossible to allocate homogeneous computing resources in the public cloud. In that case, users must use heterogeneous computing resources which have different capacity. To improve utilization in parallel database engine using computing resources on the public cloud environment, load on each node should be balanced by assigning works to nodes according to its capacity. In this paper, we propose a method which balances among nodes by assigning works to nodes corresponding its capacity, and clarify the effectiveness of the proposed method by experimental evaluation using the public cloud environment.
Journal
-
- 電子情報通信学会論文誌D 情報・システム
-
電子情報通信学会論文誌D 情報・システム J102-D (12), 773-786, 2019-12-01
The Institute of Electronics, Information and Communication Engineers
- Tweet
Keywords
Details 詳細情報について
-
- CRID
- 1390001277396826752
-
- ISSN
- 18810225
- 18804535
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
-
- Abstract License Flag
- Disallowed