International Association
for Cryptologic Research

Together with its great advantages, cloud storage brought many interesting security issues to our attention. Since 2007, with the first efficient storage integrity protocols Proofs of Retrievability (PoR) of Juels and Kaliski, and Provable Data Possession (PDP) of Ateniese et al., many researchers worked on such protocols. The first proposals worked for static or limited dynamic data, whereas later proposals enabled fully dynamic data integrity and retrievability.

Since the beginning, the difference between PDP and PoR models were greatly debated. Most notably, it was thought that dynamic PoR (DPoR) is harder than dynamic PDP (DPDP). Historically this was true: The first DPDP scheme was shown by Erway et al. in 2009, whereas the first DPoR scheme was created by Cash et al. in 2013. We show how to obtain DPoR from DPDP and PDP, together with erasure codes, making us realize that even though we did not know it, in 2009 we already could have had a DPoR solution.

We propose a general framework for constructing DPoR schemes. Our framework encapsulates all known DPoR schemes as its special cases. We further show practical and interesting optimizations that enable even better performance than Chandran et al. and Shi et al. constructions. For the first time, we show how to obtain audit bandwidth for DPoR that is independent of the data size, and how the client can greatly speed up updates with O(λ√n) local storage (where n is the number of blocks, and λ is the security parameter), which corresponds to less than 3 MB for 10 GB outsourced data, and can easily be obtained in today\'s smart phones, let alone computers.