Dynamic Online VNF Placement with Different Protection Schemes in a MEC Environment

Abstract

The Multi-access Edge Computing (MEC) architecture is made up of geographically distributed edge servers so that computing capabilities are provisioned at the network edge, close to the end users. Network Function Virtualization (NFV), when combined with MEC, provides network services in the form of Service Function Chains (SFC) with low latency. In the design of NFV-based 5G networks, the trade-off between the cost of resource deployment and the effective provisioning of services must be considered. In this work, we analyze the impact of having different MEC locations when considering the provision of SFCs in a dynamic scenario (and thus also address VNF placement). In order to deal with infrastructure failures, it is of great importance to employ robust and resilient network strategies. To safeguard SFCs against failures, various protection techniques can be applied. We use two protection methods, namely, dedicated VNF protection and shared VNF protection, under the assumption of single network failures. The operational performances of different approaches are evaluated in terms of blocking ratio and end-to-end delay, both for the whole network and for different services, and we analyze whether it is better to distribute computing servers among a few MEC sites or among a higher number.