Speaker Kevin Bowers and Tom Roeder
Host Kristin Lauter
Date recorded 5 August 2010
Cloudy, Without a Chance of Data Loss - Implementation Kevin Bowers, RSA Laboratories Cloud storage promises to provide consumers with cheap storage that can be accessed at any time, from anywhere. The appeal of such a system is easy to understand, as too is the fear people have of outsourcing the storage of valuable data. Recent failures and loss of data by cloud storage providers has done little to ease these fears. Fear is often a product of the unknown. In this talk we present several newly developed techniques that give consumers better insight into the cloud, as well as ways to leverage this insight to ensure fault tolerance. In particular, we will focus on challenges encountered while implementing Proofs of Retrievability (PORs), High Availability and Integrity Layer (HAIL), and Remote Assessments of Fault Tolerance (RAFTs). Given the right tools, the benefits of cloud storage can be achieved, without a chance of data loss. Dynamic Symmetric Searchable Encryption Tom Roeder, MSR Searchable Symmetric Encryption (SSE) allows a user to generate an encrypted index for terms extracted from a document collection. The encrypted index can be stored with the encrypted documents on a remote host, and the user can generate tokens that the host can use to search the encrypted index without revealing much information. In most applications of SSE, the encrypted documents are updated frequently, hence the encrypted index must also be updated. But published algorithms for SSE do not provide practical update mechanisms. In this talk, we introduce two new SSE schemes that provide practical update mechanisms; one scheme allows individual terms from a document to be added or deleted, and the other scheme supports adding or deleting all the terms for a document at once. We also describe a prototype implementation of the document-based scheme built over Windows Home Server.
©2010 Microsoft Corporation. All rights reserved.