Scaling application-level dynamic taint analysis to enterprise-scale distributed systems

With the increasing deployment of enterprise-scale distributedsystems, effective and practical defenses for such systems againstvarious security vulnerabilities such as sensitive data leaks areurgently needed. However, most existing solutions are limited tocentralized programs. For real-world distributed systems which areof large scales, current solutions commonly face one or more ofscalability, applicability, and portability challenges. To overcomethese challenges, we develop a novel dynamic taint analysis forenterprise-scale distributed systems. To achieve scalability, we usea multi-phase analysis strategy to reduce the overall cost. We inferimplicit dependencies via partial-ordering method events in distributed programs to address the applicability challenge. To achievegreater portability, the analysis is designed to work at an application level without customizing platforms. Empirical results haveshown promising scalability and capabilities of our approach.