A modified knowledge-based ant colony algorithm for virtual machine placement and simultaneous routing of NFV in distributed cloud architecture

The emergence of the new technologies such as virtualization and distributed cloud computing has provided new opportunities for management and orchestration of the networks by software-defined networking (SDN) and network function virtualization (NFV). SDN provides centralized knowledge about the network status and NFV lets networks implement their functions virtually on servers in edge cloud, thereby reducing costs and increasing flexibility as well as scalability of the networks. One of the main challenges for orchestration is how to increase utilization of physical network and edge cloud resources for better placement and routing of virtual network functions (VNFs) in service function chaining problem. We proposed a novel chaotic grey-wolf-optimized knowledge-based modified ant colony system algorithm, in order to have placement of VNFs and simultaneously allocate main paths and redundant paths to flows for service management by using the knowledge gained by SDN controllers. Since every flow that enters the network requires multiple virtual service functions in a service-chaining workflow, so in the proposed algorithm, service-chaining is distributed fairly on different cloudlets connected to each router in the network so that services use CPU and memory resources of all the cloudlets efficiently and fairly. We have evaluated our proposed framework by two standard network topologies connected to distributed cloudlets by realistic traffic workload. The results show that the proposed framework provides more utilization of physical resources in cloudlets by better virtual machine placement and also achieves lower delay and higher available bandwidth for VNFs in addition to better routing path redundancy. In addition, the algorithm converges faster compared to rival metaheuristic algorithms such as standard PSO for routing and placement problem.