Boundary Protection Devices (BPDs) are used by US Government mission partners to regulate the flow of information across networks of differing security levels. BPDs provide several critical functions, including preventing unauthorized sharing, sanitizing information, and preventing cyber attacks. Their application in national security and critical infrastructure environments (e.g., military missions, nuclear power plants, clean water distribution systems) calls for a comprehensive load monitoring system that provides resilience and scalability, as well as an automated and vendor neutral configuration management system that can efficiently respond to security threats at machine speed. Their design as one-way traffic control systems, however, presents challenges for dynamic load adaptation techniques that require access to application server performance metrics across network boundaries. Moreover, the structured review and approval process that regulates their configuration and use presents two significant challenges: (1) Adaptation techniques that alter the configuration of BPDs must be predictable, understandable, and pre-approved by administrators, and (2) Software can be installed on BPDs only after completing a stringent accreditation process. These challenges often lead to manual configuration management practices, which are inefficient or ineffective in many cases. The Hammerhead prototype, developed as part of the SHARC project, addresses these challenges using knowledge representation, a rule-oriented adaptation bundle format, and an extensible, open-source constraint solver.