Probabilistic Service Function Placement and Dynamic Service Function Chain Formation

Abstract

Traditional network service frameworks face significant challenges to accommodate the exponential growth in user traffic. Instead of leveraging proprietary, monolithic hardware appliances, and network functions that power up-to-day telecommunications, Virtualization will be the rightful enabling technology. Virtualization will make it possible for significant cost savings, increased flexibility and scalability in delivering new network services, and network slicing to accommodate widely differing network requirements. An ordered link of virtual network functions (VNFs) for processing traffic flows and providing end-to-end network services in response to a request is called a service function chain (SFC). To address cost reduction and performance enhancement, the primary focus should be on how to deploy VNF instances in substrate network and form SFC with the least amount of latency possible. Intelligent service function placement based on the probability of occurrence of service functions in the service function chain is incorporated in the proposed work. The service function chain is formed dynamically using the required service function. The proposed algorithm uses Dijkstra’s algorithm to find out the best service function node and shortest path between the current service function and previous one. This mechanism will minimize the number of failures in forming a service function chain, latency, and time required to form an SFC.

Keywords

• Network function virtualization
• software defined network
• service function chain
• Dijkstra’s algorithm