원문정보
초록
영어
A self-adaption fault-tolerant mechanism based on access frequency (SFMAF) is proposed for the cloud storage platform, wherein this mechanism aims at adopting the approximate LRU (Least Recently Used) algorithm to maintain a file access frequency table in order to self-adaptively adjust the fault-tolerant mode. Specifically, the replica redundancy mechanism is adopted in SFMAF for the frequently accessed files and RS (Reed-Solomon) erasure code tolerant mechanism is adopted in SFMAF for the rarely accessed files. The experiment result shows: compared with the replica redundancy mechanism, SFMAF can reduce the transmission flow of the internal system data at the cost of increasing CPU and memory usage rates in an acceptable range, namely: the system storage space is reduced.
목차
1. Introduction
2. Introduction to Hadoop Distributed File System
2.1. Hadoop Distributed File System Architecture
2.2. Fault-Tolerant Mechanism of Hadoop Distributed File System
3. Reed-Solomon (RS) Erasure Code
4. Self-Adaption Fault-Tolerant Mechanism based on Access Frequency (SFMAF) and the Implementation
4.1. Self-Adaption Fault-Tolerant Mechanism based on Access Frequency
4.2. Implementation of SFMAF Erasure code Fault-Tolerant Mechanism
4.3. Implementation of File Access Frequency Table in SFMAF
5. Experiment and Analysis
6. Conclusion
References