Data Backup Against Progressive Disasters in Geo-Distributed Data Center Networks

摘要

Geographically distributed Data Center Networks (geo-distributed DCNs) provided by large enterprises are becoming the key infrastructures to support the data stroage and data processing. However, such geo-distributed DCNs are threatened by increasing natural disasters. To this end, data backup against disasters in geo-distributed DCNs is investigated. Specifically, we focus on the scenario that DCN nodes and communication links located at different geographical regions may be threatened by a progressive disaster where early warning times are provided. An Integer Linear Program (ILP) model is formulated to back up as much data as possible within the given early waring times for the theatened DCNs. Extensive experiments are carried out to demonstrate the efficiency of the backup scheme for protecting data against a progressive disaster.