Division and Replication of Data in Cloud for Optimal Performance and Security

Abstract

Cloud technologies offer businesses a highly flexible and scalable storage infrastructure but also have significant security issues associated with their virtualized and shared nature. Traditional data protection (via encryption) guarantees confidentiality but has a high computational overhead and does not address any of the security vulnerabilities (e.g., attacks against virtual machines, nodes being compromised) of cloud computing technologies. In this paper, we present an alternative approach called "Division and Replication of Data in the Cloud for Optimal Performance and Security," or "DROPS," that will resolve these issues. Data originating from users will be divided into smaller, non-reconstructible fragments that will reside on multiple cloud nodes. Each of these fragments will then be selectively replicated (to provide high availability) so that data is protected and will not be lost if one or more nodes go offline due to failure. An unauthorized user cannot access an original file because there is no one node that can hold enough information to recreate it. Because cryptographic operations are now simpler, overall system performance is enhanced and latency time for retrieving data is decreased because these operations have been removed from the retrieval process. The system also optimally places the file fragments using the nodes' respective storage capacities, storage loads, and network bandwidths. Experimental tests in a virtualized cloud environment confirmed that the system provided better security, scalability, and availability than traditional methods.