An Eavesdropping Prevention Problem When Repairing Network Coded Data from Remote Distributed Storage

Abstract

We consider the cloud storage systems with data stored in two geographically different datacenters for remote backup. In such system, inter-data center communication is established for data repair when storage nodes fail in the data center. Since the repairing data are transmitted over the Internet, the communication between the datacenters can become susceptible to eavesdropping. This problem is especially crucial in network coding-based distributed storage systems because more repair bandwidth and repair links are required, compared to conventional replication. In this paper, we show that remote repair bandwidth can be reduced by increasing storage per node and derive the tradeoff curves between remote repair bandwidth and storage. Moreover, we show that there exist another tradeoff for storage cost and reliability for different amount of remote and local storage nodes.