Abstract

The astounding Internet revolution brings more and more new and sophisticated technologies and services, like: MPLS, VPN, QoS, RSVP, DiffServ, VLANs, bandwidth-on-demand, VoIP, etc. Moreover, the sheer number of elements in a net work is skyrocketing. For instance, an ISP may have to deal with hundreds of routers and thousands of interfaces. The diversity and heterogeneity of the network elements, domains, hierarchies, routing technologies, services and management policies gives yet another dimension to the problem. This manifold complexity poses new challenges to the network engineers and specialists. The error-prone and slow manual device configuration process involves risks like bringing the elements or the systems into undefined states or rendering them unreachable from the rest of the network and is ineffective when faced with the network's fast-growing size and heterogeneity. In this context, an integrated fabric of high- and low-level, complementary approaches is demanded, involving global- and domain-level, business policies, automated configuration, combined with outsourced policies, filtering techniques, fine-grained instance- or device-specific configuration approaches and policy error and conflict avoidance and resolution mechanisms. The report presents a configuration model which translates the manual command line information into meta-CLI constructs and allows the manipulation and composition of configurations, features, services and parameters, in order to facilitate service activation, support, invoking and monitoring, policy integration at different abstraction levels, allow better control, validation and verification, optimisation, operational efficiency and a more reliable, scalable, flexible and cost-effective configuration of the network resources and traffic