On Optimal Placement of Hybrid Service Function Chains (SFCs) of Virtual Machines and Containers in a Generic Edge-Cloud Continuum

Abstract

Traditionally, Network Function Virtualization (NFV) has been implemented to run on Virtual Machines (VMs) in form of Virtual Network Functions (VNFs). More recently, the so-called Serverless Computing has gained traction in cloud computing, offering Function-as-a-Service (FaaS) platforms that make use of containerization techniques to deploy services. In contrast to VM-based VNFs, where resources are usually reserved and continuously running, FaaS can just be subsets of code implementing small functions allowing for event-driven, on-demand instantiations. Thus, a hybrid VM-Container based Service Function Chains (SFCs) are a natural evolution of NFV architecture. We study a novel problem of optimal placement of hybrid SFCs from an Internet Service Provider (ISP) point of view, whereby VNFs can be instantiated either over VMs or containers in a generic edge and cloud continuum. To this end, we propose a Mixed-Integer Linear Programming model as well as a heuristic solution to solve this optimization problem that considers three objectives unique to the specific VM and container deployment in a carrier network: operational costs for maintaining servers in the edge, costs of placing VNFs in third-party cloud providers and penalty costs applied when SLA agreements are violated in terms of end-to-end delay. We also propose 2-phases optimization process to analyze the effect on performance as a result of replications and migrations of VNFs. The model can be used to highlight scenarios where a combination of VMs and containers can provide most benefits from the monetary costs point of view.