Available at: http://digitalcommons.calpoly.edu/theses/1819
Date of Award
MS in Computer Science
Provable Data Possession (PDP) schemes provide users with the ability to efficiently audit and verify the integrity of data stored with potentially unreliable third-parties, such as cloud storage service providers. While dozens of PDP schemes have been developed, no PDP schemes have been practically implemented with an existing cloud service. This work attempts to provide a starting point for the integration of PDP schemes with cloud storage service providers by providing a cost analysis of PDP schemes. This cost analysis is performed by implementing and analyzing five PDP schemes representative of the dozens of various PDP approaches. This paper provides analysis of the overhead and performance of each of these schemes to generate a comparable cost for each scheme using real-world cloud pricing models. Results show that the total cost of each scheme is comparable for smaller file sizes, but for larger files this cost can vary across schemes by an order of magnitude. Ultimately, the difference in cost between the simple MAC-based PDP scheme and the most "efficient" PDP scheme is negligible. While the MAC-PDP scheme may not be the most efficient, no other scheme improving upon it's complexity can be implemented without the use of additional services or APIs leading to the conclusion that the simplest, storage only PDP scheme is the most practical to implement. Furthermore, the findings in this paper suggest that, in general, PDP schemes optimize on an inaccurate cost model and that future schemes should consider the existing economic realities of cloud services.