About Virtual Waiting Time in a Multiprocessor System

Abstract

In multiprocessor queueing systems, the waiting time to start servicing tasks is crucial to optimize system performance. However, traditional measures of latency may not accurately reflect the actual task experience due to differences in service requirements and resource availability.The concept of virtual waiting time has become a valuable metric that takes into account the amount of time tasks would wait if they arrived at a given time.The article analyzes the virtual waiting time in a multiprocessor queuing system with limited queue capacity. Taking into account the distributions of arrivals and servicing of tasks, equations are derived that describe the behavior of the virtual waiting time.Based on the analysis, expressions for estimating the expected virtual waiting time are obtained. The conclusions will make it possible to optimize task scheduling and resource allocation in multiprocessor systems, increasing system performance.