Cost-Effective Reliability-Constrained Multi-Domain Service Function Chain Orchestration: A Metaheuristic Approach
Qian Zhang, Jin Yuan Sun, Yi Zhang · 2024
The orchestration of a service function chain (SFC) is responsible for the coordination of multiple virtual network functions (VNFs) in a sequence and the deployment of VNF nodes and links onto physical resources, for the purpose of delivering high-quality network services. This paper studies a multi-domain SFC orchestration problem that aims at minimizing the cost of deploying SFCs across multiple domains while ensuring the overall reliability of SFC requests. We establish a centralized multi-domain framework to formulate the SFC orchestration problem into a combinatorial optimization problem and propose a firefly algorithm (FA)-based metaheuristic approach, named as FA_SFCO, to solve the formulated problem. The proposed FA_SFCO uses a sequence of VNFs in the SFC to represent the orchestration solution, and converts all firefly individuals into valid VNF sequences to facilitate the evolutionary process of FA_SFCO. For each converted VNF sequence, FA_SFCO employs a series of operations, including backup instance allocation, SFC partitioning, and intra-domain mapping, to determine the deployment of all VNFs in the sequence, such that the fitness value of the orchestration in terms of the total deployment cost can be evaluated. Evaluation results on a multi-domain substrate network justify that FA_SFCO outperforms competing algorithms with a significant reduction in deployment cost while meeting the reliability requirement.