Venue | Category |
---|---|
HotStorage'14 | dispersal, deduplication |
Convergent Dispersal: Toward Storage-Efficient Security in a Cloud-of Clouds1. SummaryMotivation of this paperConvergent DispersalImplementation and Evaluation2. Strength (Contributions of the paper)3. Weakness (Limitations of the paper)4. Future Works
The keyless security of existing dispersal algorithms relies on embedded random information which breaks data deduplication of the dispersed data. This paper proposes convergent dispersal, which replaces the original random information with deterministic cryptographic hash which is derived from the original data.
The Summary of existing dispersal algorithm
This paper contains two convergent dispersal algorithms
CRSSS and CAONT-RS both of them augment existing dispersal algorithms with the deduplication property.
The security of existing dispersal algorithms depends on the embedded random information. Due to randomness, distinct secrets with identical content lead to different sets of shares (impede data deduplication)
measure the total amount of processed secret data divided by the computational time of generating all shares. SHA-256: for default hash AES-256: for default encryption function
In local machine
Insight: CRSSS can achieve a more flexible tradeoff between security and performance than CAONT-RS.
This paper does not analyze how the secret size affects the deduplication ratios for different dispersal algorithms. For CRSSS and CAONT-RS, since there are some constrains on the its secret sizes.