Optimization of virtual machine placement for balancing network and server load in edge computing environments

In edge computing environments, edge servers are deployed in networks in addition to cloud data centers. Designing an edge computing system involves several issues such as edge server deployment, virtual machine (VM) placement, and traffic routing. This paper focuses on the VM placement problem, which determines the location of VMs in edge servers, and proposes a new VM placement method considering network and server load. The proposed method formulates the VM placement problem as mixed integer programming (MIP) problems. By solving the MIP, the proposed method determines the optimal location of VMs so as to minimize the network load generated by the communication between users and VMs, and balance the load on edge servers. Since it takes a large amount of computation time to solve the MIP, the proposed method also introduces a heuristic algorithm for finding an approximate solution. Numerical experiments show the effectiveness of the proposed method.