Load balancing in Unihub SaaS system based on user behavior prediction

In ISP RAS cloud computing system SaaS Unihub was developed. It provides the possibility for users to work by Web-browser with interactive graphic Linux-applications, working in isolated Docker containers. Containers have dynamical demands to computational resources. Usual way of placement when containers are distributed uniformly among all servers can lead to bad result: some servers have too many applications but other are almost empty. In this paper we propose to collect information about users behavior and investigate how different applications work in order to predict containers load-balancing. Our observations show that such information can provide more uniform load-balancing and improve the whole system performance.