Efficient Reconstruction of Secret-Split Data-stores

For a while now, data reliability in distributed systems has been a major concern, which is where techniques for encryption of data were introduced. These techniques have certain flaws, like targeted theft and key management issues, when it comes to storing data for longer periods of time. Secret splitting is a method that takes care of these flaws and securely stores the data. The idea is to divide the data into multiple pieces and distributed them among multiple servers so that only authorized personels with sufficient information can access the data. The easiest way to recombine this data is to maintain a list containing information of what files go together, but the real problem occurs when this list is lost or has been compromised.
    Previous works have proposed multiple solutions for disaster recovery. Most of the ideas do extremely well for single files but are inefficient when considered for an entire datastore. We propose an algorithm called the Secret Split Secure Hash Method, which uses hints derived from same object identifiers to decide which shares go together. Although this method is not as fast as the previous methods for reconstruction of a single file, it is polynomial in time and works better when considering an entire datastore.

When:
Monday, May 1, 2017 at 1:15 PM

Where:
E2-599

CRSS Contact:
Miller, Ethan L.

Last modified 24 May 2019