원문정보
초록
영어
With the advent of the information age, the data volumes of various industries are in explosive growth, which increasingly highlights the importance of data. When a disaster occurs, how to recover the data completely and rapidly is one of the important issues receiving common concerns throughout the current industrial and academic circles. Based on the traditional disaster recovery system, this paper researches and implements a PRS algorithm-based distributed disaster recovery system[1, 2] called OsnDDR, which ensures the high availability of the system under the circumstance of continuous disasters or multi-node damage. The system adopts PRS algorithm based on RS erasure code [3] to reduce the storage resource consumption caused by the data redundancy and designs the load balancing strategy to guarantee the load balancing of each disaster recovery center in the system.
목차
1. Introduction
2. OsnDDR Architecture
2.1. System Architecture and Function of each Module
2.2. Process of OsnDRR
2.3. Design of I/O Management Module
3. Load Balancing Strategy in OsnDDR
3.1. Performance Model of the Disaster Recovery Center
3.2. Load Model of Disaster Recovery Center
3.3. Load Balancing Strategy in OsnDDR
4. Implementation Principle and Optimization of PRS Algorithm
4.1. Optimize basic Algorithm in Galois Field
4.2. Parallel Implementation of RS Erasure Code
5. Performance Evaluation
5.1. Experimental Environment Description
5.2. OsnDDR’s Impact on Application
5.3. Implement Performance Comparison of RS Erasure Code
6. Conclusions
Acknowledgement
References