Report on Policy 2006: Workshop on Policies for Distributed Systems and Networks

A policy-based system accepts as input a policy that specifies the behavior of the system in response to an event. Examples of policies include the following: (i) If a server’s average response time is below a specific threshold, then decrease the maximum number of concurrent users allowed; (ii) If there are two unsuccessful logins with the same user identifier within a three minute interval, then e-mail the system administrator. In a non policy-based system these policies would be hardcoded. Thus a change in a policy (e.g., instead of two unsuccessful logins causing an email to be sent to the system administrator have it be three unsuccessful logins) requires recoding and recompiling. In a policy-based system a change in policy results in a change of inputs to the policy-based system. This does not require recompiling, and allows for a more reusable and adaptive system. Policies have been studied in the context of areas that includes network management, quality of service and access control within an enterprise. Successes have spurred further research in exploring the use of policies in different application domains. However, commercial policy-based systems at this time are limited to VPN management and firewall management. There are a number of issues that need to be addressed in order for policybased systems to become more widely deployed. Addressing these issues is complicated by the increased complexity of computing systems as these systems become larger, and more dynamic and unstable as the use of mobile devices increases. The purpose of the IEEE International Workshop on Policies for Distributed Systems and Networks is to have a gathering of researchers to exchange