Methods for specifying and resolving security policy compliance problems

Distributed systems have become sufficiently complex that it is impractical for administrators to configure them manually to prevent security vulnerabilities. These systems consist of multiple interconnected hosts that possibly run virtualized environments and support one or more distributed applications. The administrator's job is to identify security-sensitive data and configure system components (i.e., programs, operating systems, and virtualization environments) to meet a security goal (i.e., protect data from unauthorized modification or leakage).;To prevent vulnerabilities, mandatory access controls (MAC) have been integrated into applications, operating systems, and virtualized environments. MAC systems guarantee that a system behaves within the boundaries defined by an access control policy. The problem is that although MAC systems are developed to prevent vulnerabilities, configuring several of them to work as a whole is a challenging task for system administrators. Each individual MAC policy is complex, the policies are independently developed, and the security goals that prevent vulnerabilities are usually not explicit.;We develop mostly-automated services to help administrators configure and deploy distributed MAC systems to prevent security vulnerabilities. Our results show that for commonly used deployments it is possible to use available information with little input from administrators to automate tasks that are manual currently. We reduce the burden of configuration on system administrators, thus making the deployment of MAC in distributed systems more practical.