Performance evaluation of job scheduling algorithms in P2P distributed systems

Peer-to-Peer (P2P) distributed systems have received considerable attention mainly due to their potential for high-performance and high-reliability. Job scheduling in P2P distributed systems is one of the most important issues. In this paper, we have evaluated the performance of four job scheduling algorithms named Random, Round Robin, Number of Connections and Server Load in P2P distributed computing environment. The aim of the performance evaluation is to provide a comprehensive view of what performance can be expected from these algorithms in P2P systems in terms of CPU utilization, task processing, traffic received and traffic sent at different load scenarios. We used OPNET modeler to build the simulation model for FTP application on the P2P distributed systems. The output of the performance comparison illustrates different QoS (Quality of Service) aspects of these basic algorithms under different workloads and traffic settings. The results show that the performance of each algorithm changes according to the operational environment setting. The simulation results show that the Server Load-based scheduling algorithm has the best performance among the four algorithms followed by the Number of Connections-based scheduling algorithm. Simulations also show that the Server Load scheduling policy is not suitable for heavy load situations and the Random algorithm is suitable for some settings, especially in the low load situation.