Network Performance and Quality of Experience of Remote Access Laboratories

Abstract

Remote Access Laboratories (RAL) have become important learning and teaching tools. This paper presents a performance study that targets a specific remote access architecture implemented within a universities operational environment. This particular RAL system provides globally authenticated and arbitrated remote access to virtualized computers as well as computer controlled hardware experiments. This paper presents system performance results that have been obtained utilizing both a set of automated and human subject tests. Principle objectives of the study were: To gain a better understanding of the nature of network traffic caused by experimental activity usage; to obtain an indication of user expectations of activity performance; and to develop a measure to predict Quality of Experience, based on easily measurable Quality of Service parameters. The study emulates network layer variation of access-bandwidth and round-trip-time of typical usage scenarios and contrasts against user perception results that allow classifying expected user performance. It demonstrates that failure rate is excellent measure of usability, and that round-trip-time predominantly affects user experience. Thin-client and remote desktop architectures are popular to separate the location of users and the actual data processing and use similar structures, hence results of this study to be applied in these application areas as well.