Policies for allocating performance impairment budgets among multiple IP providers

Abstract

An important technical aspect of achieving end-to-end quality of service (QoS) in next generation networks refers to allocation of the total performance impairment budget (delay, jitter, packet loss) among multiple providers. In this article, we first propose a generic policy for allocating per-domain impairment budgets, relying on the set of performance metrics from service request and the rules for their composition on the multi-domain path. The objective is to provide end-to-end QoS through the set of heterogeneous domains, with different QoS models and definitions of service classes. The allocation of impairment budgets among multiple domains is then closely related to mapping of service classes between providers and selection of the most appropriate class for particular service in each domain. Based on the generic policy, we further derive examples of specific policies and evaluate them with respect to fulfillment of QoS objectives, fairness, adaptability and scalability. Evaluation tool implements a policy-based conformance matching scheme, which enforces selection of the domain class that most tightly matches with the required QoS.