Abstract

Ken Birman: Scalable Trust

We're starting to see a serious effort by large enterprises to build trust management systems that can span huge numbers of machines and users.  Conceptually, such a system consists of a central database of security policy rules and a great number of agents charged with intercepting operations controlled by these policies, determining if the policy is satisfied, and then allowing or disallowing the policy as appropriate.  But scaling a trust management system up introduces additional problems.  This talk will focus on the kinds of new questions raised by scalable trust management platforms, asking to what extent things we know from past work can be used to develop these systems, and identifying a number of open questions on which (much) more research is needed.  In particular, we'll ask what roles Cornell's new Quicksilver platform might play in a scalable trust management system, and how Quicksilver might need to be extended were it to be used in this way.