Fault Tolerance for Parallel Applications through Replication

• المؤلف الرئيسي: SHUM, Kam Hong
• التنسيق: text
• اللغة: English
• منشور في: Institutional Knowledge at Singapore Management University 1997
• الموضوعات: Databases and Information Systems; Numerical Analysis and Scientific Computing
• الوصول للمادة أونلاين: https://ink.library.smu.edu.sg/sis_research/1054; http://dx.doi.org/10.1109/ICICS.1997.652234
• المؤسسة: Singapore Management University

Based on the technique of replication, an efficient fault-tolerant model for parallel computing on workstation clusters is proposed. The model is built on top of a runtime system which supports resource allocation for parallel applications running on heterogeneous workstation clusters. According to the results of resource allocation, replicated parallel applications can minimize their resource consumption by runtime reconfiguration. Besides, checkpointed states only transfer among replicated applications, no expensive disk read/write operations are therefore required.