Quantitative Service Differentiation: A Square-Root Proportional Model

Abstract

AbstractDue to the open and dynamics nature of ubiquitous computing environments and services, quantitative service differentiation is needed to provide controllable quality of service (QoS) levels to meet changing system configuration and resource availability and to satisfy different requirements of applications and users. A proportional differentiation model was proposed in the DiffServ context, which states that QoS factors of certain classes of aggregated traffic be proportional to their differentiation weights. While it provides compelling proportionality fairness to clients, it lacks of the support of a server-side QoS optimization with respect to the resource allocation. In this paper, we propose and promote a square-root proportional differentiation model for delay-sensitive Internet services. Interestingly, both popular QoS factors with respect to delay, queueing delay and slowdown, are reciprocally proportional to the allocated resource usages. We formulate the problem of quantitative service differentiation as a resource allocation optimization towards the minimization of system delay, defined as the sum of weighted responsiveness of client request classes. We prove that the optimization-based resource allocation scheme essentially provides square-root proportional service differentiation to clients. We then propose a generalized rate-based resource allocation approach. Simulation results demonstrate that the approach provides quantitative service differentiation at a minimum cost of system delay.KeywordsResource AllocationAdmission ControlSystem DelayRate AllocationService Time DistributionThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.