Improving Resource Efficiency in Internet Cafés by Virtualization and Optimal User Allocation

Abstract

The concept of using distributed computing to supply video games to end users has been growing in popularity. Internet cafés are one potential application for this concept. We consider a cloud-based model for Internet cafés where servers provide virtual machines with different specifications in order to meet different kinds of user demand (web browsing, low end gaming, medium end gaming, and high end gaming). In an Internet café, users arrive throughout the day with different demands and different durations for which they stay. Given the user demand over time and a fixed hardware set-up of servers, the task then consists of choosing which users to accept and how to allocate the accepted users to the servers in order to maximize the total profit of the Internet café. We formulate an integer programming model for computing an optimal choice of users to accept together with an efficient allocation of accepted users to servers. Computational results show that, when allocating users efficiently, using a cloud-based setting with servers providing virtual machines that exactly meet the users' demands can greatly improve resource efficiency in Internet cafés compared to classical zoning models that use desktop computers. At the same time, the total profit obtained from accepting users can be improved significantly due to the added flexibility when using an optimized user acceptance strategy.