Blindfold: a system to "See no evil" in content discovery

  • Authors:
  • Ryan S. Peterson;Bernard Wong;Emin Gün Sirer

  • Affiliations:
  • Department of Computer Science, Cornell University, United Networks, L.L.C.;Department of Computer Science, Cornell University, United Networks, L.L.C.;Department of Computer Science, Cornell University, United Networks, L.L.C.

  • Venue:
  • IPTPS'10 Proceedings of the 9th international conference on Peer-to-peer systems
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Existing content aggregators provide fast and efficient access to large volumes of shared data and serve as critical centralized components of many peer-to-peer systems, including content discovery for BitTorrent. These aggregators' operators are tasked to spend significant human resources to manually vet uploaded data to ensure compliance with copyright laws. This task does not scale with today's increasing demand for such services. In this paper, we introduce Blindfold, a scheme to ensure that the operators of content aggregators are completely blind to the content that they are storing and serving, thereby eliminating the possibility to censor content at the servers. It works by partitioning the search and upload operations into a series of dependent key-value operations across servers under different administrative domains, with the connection between servers obfuscated using captchas. We have implemented a prototype of Blindfold to show that it is a simple, feasible, and efficient system for serving content that is opaque to the storage servers.