An Evaluation on Power Consumption and Performance Balancing Distributed Storage Systems
スポンサーリンク
概要
- 論文の詳細を見る
In distributed storage system, replicating data on multiple nodes for improving performance also extends power consumption of whole system. As the result, it becomes more important to control both power consumption and performance in storage system. Recently, a number of proposals that allows the system to be available to control power consumption have been proposed and confirmed to be success in certain circumstance and applications. In these proposals, system is made available to operate in multiple modes with containing different number of active datanodes that store data. Consequently, at each mode, the system obtains performance which is proportional with the number of active datanodes it contains. However, this approach faces a problem of performance degradation when the system needs to correct its data layout while moving up to high mode from power-saving modes if the dataset is updated during power-saving mode. In this paper, the function of data layout correction is added into two existing data placement methods in distributed storage system and the degradation in performance of these methods is then evaluated through empirical experiments.
- 2011-07-26
著者
-
Haruo Yokota
Department Of Computer Science Graduate School Of Information Science And Engineering Tokyo Institut
-
Yokota Haruo
Department Of Computer Science Graduate School Of Information Science And Engineering Tokyo Institut
-
Yokota Haruo
Tokyo Inst. Of Technol.
-
Haruo Yokota
Department Of Computer Science Graduate School Of Information Engineering And Science Tokyo Institut
-
Satoshi Hikida
Department of Computer Science, Graduate School of Information Engineering and Science, Tokyo Instit
-
Satoshi Hikida
Department Of Computer Science Graduate School Of Information Engineering And Science Tokyo Institut
関連論文
- MARK-OPT: A Concurrency Control Protocol for Parallel B-Tree Structures to Reduce the Cost of SMOs(Database)
- Comparing Hadoop and Fat-Btree based access method for Small File I/O Applications
- Comparing Hadoop and Fat-Btree based access method for Small File I/O Applications
- 自律ディスクにおける任意サイズのデータの取り扱い
- An Evaluation on Power Consumption and Performance Balancing Distributed Storage Systems
- An Evaluation on Power Consumption and Performance Balancing Distributed Storage Systems
- Parasporin-1, a Novel Cytotoxic Protein to Human Cells from Non-Insecticidal Parasporal Inclusions of Bacillus thuringiensis
- A Compound Parallel Btree for High Scalability and Availability on Chained Declustering Parallel Systems
- XML Documents Searching Combining Structure and Keywords Similarities