Diagnosing configuration errors in virtual private networks

Traditional network fault management systems diagnose hard, localized errors such as fiber cuts or hardware/software component failures. It is quite possible, however, that network components work correctly yet end-to-end services do not. This happens if there are configuration errors, i.e., configuration parameters of components are set to incorrect values. Configuration is a fundamental operation for integrating components to implement end-to-end services. Configuration errors are frequent because transforming end-to-end service requirements into configurations is inherently difficult: in realistic networks there are many components, configuration parameters, values, protocols and requirements. Yet, such transformation is largely, manually performed. This paper describes a toolkit called Service Grammar to diagnose configuration errors. The toolkit is illustrated in the context of an IP virtual private network with routing and security services. It is based on the following assumptions: (1) every component has configuration parameters which can be set to definite values; (2) these values remain fixed during the normal operation of a component; and (3) the set of values of all configuration parameters in a network, called the configuration vector, determines the behavior of the network as a whole