The Influence of Workload Characterization in the Performance Analysis of Tele-collaboration Environments V. Mahadevan R. Braun Z. Chaczko mahadevn@eng.uts.edu.au Robin.Braun@uts.edu.au zenon@eng.uts.edu.au ICT Group Faculty of Engineering University of Technology Sydney NSW 2007 Australia Abstract- The Telecollaboration (TC) business system, an important class of emerging applications that is spawning new challenges to achieving the desired levels of performance. One important implication in identifying and characterizing the properties of TC system implementations is that performance evaluation should no longer be isolated and viewed as a separate analytic activity. Instead, we require a physical and logical understanding of the complex quality issues that affect overall performance of applications, systems and network infrastructures. As one of the natural artifacts of this practice, a workload model will be developed to characterize a TC system. We demonstrate how such an analysis may result in characterization of workload behavior, and that in turn will lead to definition of efficient analytical model parameters, network invariants and use of relevant quality metrics. The study of underlying issues of Telecollaboration Quality of Service (TQoS) within the context of TC could have positive impacts on performance analysis and traffic control, if the analytical modeling used remains consistent and effective in describing the Web-based environments. I. BACKGROUND Over the last decade, the emergence of seamless integration of text, sound and images with the numerous benefits of collaboration tools have served as a new generation platform for the development and deployment of innovative IP-based applications and services. The CeNTIE’s (Centre for Networking Technologies for the Information Economy) TC project at UTS (University of Technology Sydney) aims to target to the enterprise market, as part of other major ICT (Information and Communication Technologies) initiatives that are currently being undertaken Australia wide [4, 5]. Due to increases in their user volumes, these applications might be subject to failures due to unscheduled maintenance, human errors and intermittent system outages during their usage. This in turn will have a huge impact on collaborative business outcomes of the participants of TC business system with the significantly degraded performance. Ultimately such scenarios could lead to loss of business. Daniel A. Menascé and Virgilio A. F. Almeida showed that quantitative descriptions of system behavior plays a major role in terms of understanding, analyzing, designing and operating Internet- based large-scale systems such as TC. That poses new challenges for the rapid explosion of scale and demand for QoS dependant systems [1]. But for TC, there are other collaborative business objectives that need to be met, not just to increase the level of their employee productivity but to improve employee decision-making and problem solving capabilities for the enhancement of the participant’s collaborative outcome experience. In this paper we first emphasize the evolution of TC business systems along with the identification of its client/server/LAN (Local Area Network) based systems, resources and workload parameters. This includes the introduction of quality metrics involved in workload characterization of TC web sites through quantitative approaches that are required. An introduction of models involved in a SCM (Supply Chain Management) TC scenario will follow, that would assist us to capture the characterization of the complex interaction between the participants and TC web sites. This would even enable the participants to identify the QoS metrics of TC environments which will then become a basis for the formative and summative performance evaluation of invariants of Internet workloads. The second part of this paper insists on the importance of in-depth understanding of workload characterization and performance evaluation of Internet traffic for collaborative interactions of participants through use of analytical studies. The rest of the paper outlines the necessity of a theoretical and practical understanding of the influence of workload characterization on performance evaluation study of this innovative application, and speculates on likely future developments in these two influential areas. II. EVOLUTION OF TC BUSINESS SYSTEM For analysis purposes, authors have decided to take a SCM system [9] to contribute QoB (Quality of Business) metrics that allows B2B (Business-to-Business) transactions and collaboration activities between groups of people in delivering a set of services at a level of enterprise applications and collaborative business processes. The participants in this scenario are identified as businesses and their suppliers/trading partners. Fig. 1 shows a model of the collaborative SCM system that employs systems such as workflow management, document management, discussion and videoconferencing [7].