Improving Usability and Security in the Semantics of System Configuration Languages

Increasing complexity and demand of IT services has led to a large strain on system administrators. System configuration languages are intended to make their job easier by leveraging a single, consistent language in which they can express desired system states for a large variety of computers. However, despite the existence of such languages, configuration errors are still prevalent in production systems and frequently result in disruptions to availability and security vulnerabilities. This is largely due to misunderstanding of the languages and related concepts by system administrators. In this work I look at the semantics of configuration languages and how users interpret and mentally parse these languages. The aim of this research is to better understand the problems with existing configuration languages and to gather data which will lead to the design of more usable languages. This in turn should result in fewer configuration errors, large outages and security vulnerabilities.