Candidate
Jeff Yucong Luo
Title
Secure Cloud Storage
Supervisor
Abstract
Since the early 2000s, the use of the Cloud as a remote backup storage system has been gaining popularity. However, it has been marred by a continuous series of critical security attacks resulting in the access, transfer, and loss of many customer data and corporate secrets. Internet users are at the whims of Cloud storage providers in terms of the security and privacy protections provided for their data, as well as the service availability and pricing. At the same time, there are numerous advantages to storing data on the Cloud. This includes being able to access the data worldwide, recover from local equipment failures and data losses, reduce management and maintenance costs, synchronize and share files to teams, and improve global resource utilization.
Users currently must make a conscious choice of accepting and facing these drawbacks in order to reap the benefits of storing data on the Cloud. Tackling this trade-off is the central problem for Secure Cloud Storage. As a secondary objective, minimizing the total storage costs would be ideal. Solving this problem would significantly improve the security and reliability of data stored on the Cloud, and consumer confidence of using the Cloud.
To solve this problem, a software system was designed based on the use of Erasure Codes on the user's computer. The system can split users' files into pieces and add redundancy to the pieces to tolerate data loss. In reverse, the system can recover an original file using only a subset of the file pieces. The file pieces are sent independently to multiple Cloud storage providers, so that attacks endured or mistakes made by a few providers will not affect the ability to recover the file for the user. This approach also maintains the benefits, and resolves the economic challenges of using the Cloud as a storage system.
In addition to presenting the advantages and drawbacks of the current Cloud storage systems, the seminar will provide a qualitative analysis of how the proposed approach resolves the Cloud storage problem while maintaining its benefits, and an overview of the software system design. A comparative analysis of seven erasure codes applicable in this problem space will also be presented to demonstrate the diversity of choices and the performances of the different erasure codes in the context of storage.