Constraint-based specifications for system configuration

Abstract

Declarative, object-oriented configuration management systems are widely used, and there is a desire to extend such systems with automated analysis and decision-making. This thesis introduces a new formulation for configuration management problems based on the tools and techniques of constraint programming, which enables automated decision-making. We present ConfSolve, an object-oriented declarative configuration language, in which logical constraints on a system can be specified. Verification, impact analysis, and the generation of valid configurations can then be performed. This is achieved via translation to the MiniZinc constraint programming language, which is in turn solved via the Gecode constraint solver. We formally define the syntax, type system, and semantics of ConfSolve, in order to provide it with a rigorous foundation. Additionally we show that our implementation outperforms previous work, which utilised an SMT solver, while adding new features such as optimisation. We next develop an extension of the ConfSolve language, which facilitates not only one-off configuration tasks, but also subsequent re-configurations in which the previous state of the system is taken into account. In a practical setting one does not wish for a re-configuration to deviate too far from the existing state, unless the benefits are substantial. Re-configuration is of crucial importance if automated configuration systems are to gain industry adoption. We present a novel approach to incorporating state-change into ConfSolve while remaining declarative and providing acceptable performance.