An Analytical Model to Design Processor Sharing for SDN/NFV Nodes

摘要

The introduction of the two new paradigms Software Defined Networks (SDN) and Network Functions Virtualization (NFV) in the Internet is revolutionizing its design and management, creating many new challenging problems regarding resource allocation and orchestration. Little attention has been paid up to now to the "intranode" resource allocation. In this context, this paper introduces an analytical model of an SDN/NFV node that implements NARR, a processor sharing strategy proposed by the same authors to reduce delays by assigning a larger processor portion to the flows that will leave the node through a less loaded output network interface card (NIC). Numerical results analytically demonstrates the gain achieved by applying NARR, and give important insights for future work.