Verifiable Data Redundancy in the Cloud
M. Kamrul Islam, Ragib Hasan · 2016
Data redundancy is critical for the assurance of service continuity in fault tolerant systems. Researchers have proposed several efficient data replication strategies to secure users' sensitive information from a single point of failure. Although the cloud service providers (CSP) assure users regarding data safety by following some of those replication strategies, the opaque operational model of the cloud does not allow the users to verify the replications. This, in turn, introduces a lack of trust in clouds. The challenge of verifiable data redundancy is two-fold: to verify that the server indeed possesses multiple copies, and to verify that those copies are not located on the same physical system. Current research mainly focuses on verification by access time of multiple requests which is highly system dependent. Hence, we propose a novel storage-as-a-service protocol that provides verifiable data redundancy in the semi or untrusted server. Our proposed model is completely oblivious of the physical system and the allocation protocol of the cloud. It does not include the server in the verification process which eliminates the risk of manipulation by a dishonest service provider. Our model uses distinct copies to store in the server as replicas and provides deterministic verification of having data redundancies in the server. For generating distinct copies of data, we use different random numbers in the ElGamal encryption system. After proposing the storage, access, and verification processes, we implemented a prototype system to analyze the performance of our proposed model. Our experimental results show that our system can successfully verify the redundancy of the data in the remote server and detect any service level agreement anomalies.