A Direct Data Fetch technology applied in disaster-tolerant model of Distributed File System

A geographically isolated disaster-tolerant system for Distributed File System (DFS) is essential for its enhancement in data security. However, the plain disaster tolerant model performs poor quality while a switch happens in hot-backup. In this study, we proposed a Direct Data Fetch (DDF) technology for the hybrid model. The DDF can fetch data directly from the data chunk server while the main server is temporarily unavailable. Then, we implemented the model on MooseFS, and compared it with IBM Peer-to-Peer Remote Copy (PPRC) technology. The test results indicate that DDF technology enhanced the stability and performance during the switch process of disaster tolerant model, and performs best in small quantities of files.