Grid Scheduling of Super-Peer Model Based on Price Timed Petri Net

For grid of super-peer model where nodes were clustering together by means of semantic similarity, a compound task scheduling algorithm was proposed which taking users' time deadline, cost limit and the weight parameters between them as QoS requirements. Considering grid nodes were dynamic, the rescheduling mechanism was introduced. Price timed Petri net which was a powerful graphical and mathematical tool was used to formally and hierarchically model grid scheduling, in order to have better understanding about it. Reachable task graph of Petri net model was constructed to analyze scheduling performance, such as scheduling time, cost, throughout capacity and load balance, etc. Effectiveness of the scheduling algorithm was verified by an example, scheduling process of compound tasks which need several types of resources and the system's scheduling performances were analyzed. Related work had great significance on guiding the construction and implement of real grid systems of super-peer model.