Optimal Placement of VNFs and Service Function Chaining in an Edge Computing Environment
Abhay Shirol, Vijayalakshmi M, Sanjana Dyavappanavar, R. H. Shrinidhi · 2024
Network Function Virtualization (NFV) and Service Function Chaining (SFC) are developing concepts that are gaining acceptance in modern networking. These paradigms were developed in response to the increasing complexity of network services and the need to create more efficient, adaptable, and scalable methods of delivering these services. The chaining of network services across several domains is made possible by SFC as an extension of NFV. SFC entails the sequential grouping of network functions to form a chain of functions that are applied to incoming packets. This allows network managers to customize network services and implement sophisticated capabilities like deep packet inspection, load balancing, and firewall in a scalable and efficient manner. SFC, on the other hand, is an NFC extension that enables the chaining of network services over various domains. This enables network operators to design advanced networks, without having to rely on complex and inefficient manual configuration. SFC has been adopted in a wide range of applications and use cases, including cloud computing, Internet of Things (IoT), and software-defined networking (SDN). These paradigms have the potential to provide more agile and flexible network architectures. For obtaining efficient chaining that takes the minimum latency delay possible to service the virtual functions, we have used Q-Learning Method Reinforcement based machine learning. We have also considered the case in which multiple VNFs are present in MECs. By implementing this we obtained that using Q-Learning, we can keep the state of previously chosen MEC for the same SFC so that the same optimal path will be selected. Additionally, we can see that the graph for computing time for the SFC request, which has a longer VNF, won’t show a steep increase.