Check out the new USENIX Web site. next up previous
Next: Scalability Up: DisCFS Design Previous: Security Analysis

Threat Analysis

  At the object level, the threat model of DisCFS does not seem any different from any simple file access control protocol. DisCFS does not encrypt files on disk, thus users have the option of trusting the system administrator or using encryption mechanisms on top of DisCFS.

At first glance the threat model of DisCFS at the network level does not differ from that of NFS used over a secure channel, e.g. a trusted LAN or VPN. However, what sets DisCFS apart is its ability to address the threat of implicit rights amplification inherent in identity-based access control. All co-authors writing an arbitrary paper using CVS need to have login access to the serving machine. In contrast, the credential-based access control of DisCFS allows us to trust the co-authors no more than we have to - for authorship and nothing else.

Stefan Miltchev