Models of dynamic relations among service activities, system state and service quality on computer and network systems

摘要

Service quality on computer and network systems has become increasingly important as many conventional service transactions are moved online. Service quality of computer and network services can be measured by the performance of the service process in throughput, delay, and so on. On a computer and network system, competing service requests of users and associated service activities change the state of limited system resources which in turn affects the achieved service quality. Modeling dynamic relations of service activities, system state and service quality is required to determine if users' service requests and requirements of service quality can be satisfied by the system with limited resources and how the system and service configuration can be adapted to meet service quality requirements. This paper presents our empirical study to establish activity-state-quality models for a voice communication service. We run experiments to collect system dynamics data under various service conditions and use statistical techniques to analyze experimental data and build activity-state-quality models. The results reveal four major types of dynamic relations among service activity parameters, system resource state, and the network throughput - a measure of achieved service quality for the voice communication service. Although delay-related measures are also important for voice data communication, they are not collected in this study. Five system state variables concerning the memory, CPU, process and IP resources are uncovered to be affected by service activity parameters significantly and be associated with the achieved service quality closely. We also obtain an insight about increasing the size of the buffer which holds voice data before transmission over the network to alleviate the workload on system resources and maintain the network throughout when the number of client requests and the client requirement in voice quality increase.