Computational and Network Utilization in the Application of Thin Clients in Cloud-Based Virtual Applications

Abstract

Typically, training using virtual environments uses a client-server or a fully distributed approach. In either arrangement, the clients used are full computers (PCs) with an adequate processor, memory, and graphics capability. These are reasonably costly, require maintenance, and have security concerns. In the office desktop environment, the use of thin clients is well known; however, the application of thin clients with cloud-based servers to virtual training is relatively new. Thin clients require less initial cost, require less setup and maintenance, and centralize the virtual environment configuration, maintenance, and security to virtual cloud servers. Rather than housing an expensive computer (a so-called thick client) at each station, functionality is replaced using a streaming protocol, a remote server, and a thin client to allow the user to interact. This paper reviews two game-focused streaming protocols running across a set of four thin clients (of various capability and cost) from both local and remote cloud-based data centers. Data were gathered to measure latency and network and computational utilization across each client using two scenarios in both local and remote conditions. Results of these experiments indicate that thin clients for use in virtual training is viable regardless of local or remote server location.