Adaptive Network Coding Based Disaster Backup in Cloud-of-Clouds Storage Systems

Cloud-of-clouds storage provides a practical solution to enhance security and reliability of distributed data storage, by encrypting, encoding and storing data across multiple clouds. However, it is a great challenge to adopt such a paradigm in disaster backup applications, which need to transmit large data volume with limited network bandwidth. To this end, we improve the performance of the backup applications in cloud-of-clouds storage systems by allowing backup clouds to perform encoding and transmission operations during data storage process, so that the storage systems can adaptively adjust the network traffic between clouds. By proposing a Network Coding based Cloud-of-Clouds Backup (NC
³
B) scheme to encode and exchange data collaboratively, we develop two state-of-the-art Reed-Solomon (RS) based data storage approaches, i.e., All-Or-Nothing Transform-RS (AONT-RS) and Converge AONT-RS (CAONT-RS), into Network coding based Backup AONT-RS (NBAONT-RS) and Network coding based Backup CAONT-RS (NBCAONT-RS), respectively. We conduct extensive experiments in Amazon EC2 with a real-world prototype storage system and cluster trace set. The experimental results show that the developed approaches can reduce the latency of data storage process by up to 50% and improve the throughput by up to 98% compared with their state-of-the-art counterparts.