Latency-based graph selection manager for end-to-end network service on heterogeneous infrastructures

Abstract

As Network Functions Virtualization (NFV) / Software Defined Networking (SDN) technologies gradually mature as representative next-generation network technologies, management and orchestration (MANO) technologies that manage the Service Function Chain (SFC) have received plenty of attention from industry and academia. However, a number of present research challenges have been addressing only resource efficiency related with usage rates of Virtualized Network Functions (VNFs) without network properties in terms of NFV-MANO orchestrator. Different network characteristics are considered due to use network slicing technology which provides dedicated logical networks from a physical network for tailored network services. In this paper, we proposed Graph Selection Manager (GSM) to provide one or more VNF forwarding graphs under maximum latency considering inter-latency and intra-latency as well as VNFs and network capacity. As a result of experimental emulation, we found one or more VNF forwarding graphs meeting service level agreement (SLA) of a tenant in limited network capacity, and established one or more end-to-end network services with sustained low latency.