Methods for Objective and Subjective Evaluation of Zero-Client Computing

Abstract

Zero clients are hardware-based devices without a central processing unit (CPU) that deliver virtual desktops (VDs) from remote computing systems to users. We measured the performance of applications accessed through zero clients to study the feasibility of using this approach to provide a desktop-pc experience across a network. Performance evaluation is complicated because monitoring software cannot be downloaded to the zero clients. Therefore, we introduce a new methodology and metric to measure zero-client VD performance that is based on network-traffic analysis. We conducted objective and subjective studies to determine the sensitivity of application-specific metrics to different network conditions. The results show that the packet loss rate (PLR) impacts zero-client performance for some applications such as video streaming. Subjective tests showed a greater user sensitivity to the PLR for video streaming than for image viewing or Skype. A strong correlation was found between the objective and subjective measurements but the rate at which these measurements changed with increasing PLR differed depending on the application.