A Distributed Policy-based Network Management (PBNM) system for Enriched Experience NetworksTM (EENs)

Policy-based Network Management (PBNM) systems have traditionally been used in enterprise networks where the emphasis has been on centralising the responsibility for management of the network for simplicity. In this environment, PBNM systems have been helpful in partially automating the configuration of the network and prioritising the use of resources. It is now being considered that PBNM systems might be useful in carrier-class networks for the delivery of the next generation of broadband services to both residential and business customers. This has broadened the focus to simplifying and automating the configuration and monitoring of more complex services across heterogenous networks. As such, this creates a new form of broadband network, called an Enriched Experience NetworkTM (EEN), which requires an adaptable approach to service management that allows for personalised and ubiquitous services which are available on-demand. This is necessary to enrich the user’s experience of the network services. By making the services tailored to more specific needs, their value to the users is increased. PBNM systems are promising here because they have to potential to abstract concepts to a customer or service management level whilst managing the complexities of service delivery. In carrier-class networks, there is a greater emphasis on the reliability, performance and scalability of the management systems, because of the need for “five nines” availability of services to particular customers. A broadband service delivery framework must support up to millions of users, services and devices simultaneously. This suggests that a distributed architecture is required to increase both the throughput and redundancy of the PBNM system so that the non-functional objectives can be met. Crucial to a distributed PBNM system, there will be a number of management entities that have the responsibility for managing separate and independent partitions of the network. Whilst some policies will be simple to compartmentalise and distribute, other policies will describe network behaviour that crosses a number of independent partitions. A methodology is required to coordinate the behaviour of these autonomous management entities, to ensure that overall service objectives are being met.