Service function chain embedding in centralized and distributed data centers - A comparison

Cloud computing has played an important role in providing IoT-based services recently, such as healthcare, smart grid, driving-assistant systems and so forth. In such a paradigm, there is a tendency to deploy services in the edge-cloud environment, where data centers or computing clusters are partly moved to the edge of the network to avoid service degradation or disruption due to the scarcity of physical resources. This paper analyzes and discusses the advantages and disadvantages of providing virtualized services based on network function virtualization in two edge-cloud scenarios, in which data centers are in the center or placed at the edge of the network. Furthermore, a novel service function chain embedding strategy has been proposed, which considers centralized or multiple distributed DCs scenarios, and focuses on DC-internal embedding in fat-tree fabrics under online arrivals and resource fragmentation. Performance evaluation results show that the proposed strategy can improve the efficiency of the cloud system in terms of resource utilization and power consumption.