Estimation of the required number of nodes of a university cloud virtualization cluster

Abstract

When designing a virtual desktop infrastructure (VDI) for a university or inter-university cloud, developers must overcome many complex technical challenges. One of these tasks is estimating the required number of virtualization cluster nodes. Such nodes host virtual machines for users. These virtual machines can be used by students and teachers to complete academic assignments or research work. Another task that arises in the VDI design process is the problem of algorithmizing the placement of virtual machines in a computer network. In this case, optimal placement of virtual machines will reduce the number of computer nodes without affecting functionality. And this, ultimately, helps to reduce the cost of such a solution, which is important for educational institutions. The article proposes a model for estimating the required number of virtualization cluster nodes. The proposed model is based on a combined approach, which involves jointly solving the problem of optimal packaging and finding the configuration of server platforms of a private university cloud using a genetic algorithm. The model introduced in this research is universal. It can be used in the design of university cloud systems for different purposes-for example, educational systems or inter-university scientific laboratory management systems.