A Practical System for Verifying Recoverability of Big Data
Date: Wednesday, December 11, 2013
Time: 4:00 PM to 5:00 PM Note: all times are in the Eastern Time Zone
Refreshments: 4:00 PM
Host: Prof. Srini Devadas
Contact: Raluca Ada Popa, firstname.lastname@example.org
Speaker URL: None
TALK: A Practical System for Verifying Recoverability of Big Data
Abstract: We designed and built a practical and efficient dynamic PoR (proof of retrievability) system that allows a client to quickly verify that an untrusted or compromised cloud is storing all of the client's data and that none of the data is corrupted beyond repair. Unlike previous work, our system requires very little client storage and uses about 300 times less client-server bandwidth, resulting in almost no bandwidth overhead. Our system is capable of verifying terabytes of data in less than 9 seconds, and can easily scale (logarithmically) to much larger storage systems.
We achieve this by introducing several powerful techniques, including a novel incrementally-constructible erasure code based on the Discrete Fourier Transform (DFT). Because the techniques we proposed are extremely scalable and efficient, corporations can use them to ensure the reliability of all (possibly petabytes) of the information they store in the cloud. Prior to our work, it was prohibitively impractical to verifiably impose this level of reliability at a large scale.
Created by Raluca Ada Popa at Wednesday, December 11, 2013 at 1:29 AM.