Breaking the Limits: Bio-Inspired SFC Deployment across Multiple Domains, Clouds and Orchestrators

The Multi-SFC strategy enables the deployment of virtual network services across multiple clouds, domains, and NFV orchestrators. In this context, finding optimal solutions to the NFV Resource Allocation (NFV-RA) problem, and in particular the VNF Forwarding Graph Embedding (VNF-FGE) problem, becomes a formidable challenge. This translates into a multi-objective optimization problem that involves multiple heterogeneous resources, services and functions, each with particular sets of features and restrictions. In this work, we propose a bioinspired strategy for mapping Multi-SFCs, which consists of a multi-objective genetic algorithm that minimizes inter-domain latency and resource usage costs, while maximizing inter-domain bandwidth availability. Furthermore, the solution is dynamic in the sense that service re-mapping can be invoked after network conditions change. Simulation results show the effectiveness of the proposal to create and adapt optimized solutions.