A queuing network model for minimizing the total makespan of computational grids

Performance aware scheduling considering resource availability in grid computing

“…Both QN and GSPN models proposed in [9] only consider grid tasks submitted by grid users paying no attention to the local tasks of the system, whereas the models proposed in [8] consider both grid and local tasks. Difficulty with all of the models proposed in [8] and [9] is that the models can only compute pure performance measures of the grid environment, and they do not handle the situation in which one or more of the resources fail. Azgomi et al [3] have presented a Colored Petri Net (CPN) model to show the workflow of task execution in grid computing, and compute the reliability of a grid service.…”

“…Parsa et al [8,9] have proposed Queuing Network (QN) and Generalized Stochastic Petri Net (GSPN) solutions to model and evaluate the performance of grid computing systems. The performance measures evaluated in [8] and [9] are the total makespan of the environment and the mean number of grid tasks waiting to receive service from the environment, respectively.…”

“…The performance measures evaluated in [8] and [9] are the total makespan of the environment and the mean number of grid tasks waiting to receive service from the environment, respectively. Both QN and GSPN models proposed in [9] only consider grid tasks submitted by grid users paying no attention to the local tasks of the system, whereas the models proposed in [8] consider both grid and local tasks.…”

“…The performance measures evaluated in [8] and [9] are the total makespan of the environment and the mean number of grid tasks waiting to receive service from the environment, respectively. Both QN and GSPN models proposed in [9] only consider grid tasks submitted by grid users paying no attention to the local tasks of the system, whereas the models proposed in [8] consider both grid and local tasks. Difficulty with all of the models proposed in [8] and [9] is that the models can only compute pure performance measures of the grid environment, and they do not handle the situation in which one or more of the resources fail.…”

“…Using the makespans of all grid resources computed in the previous step, the total makespan of the entire grid environment can be computed. Hence, a scheduling algorithm based on the Simulated Annealing (SA) metaheuristic is presented to find a good enough scheduling of jobs on resources with the aim of minimizing the total several related measures such as the total completion time of all jobs, expected waiting time of jobs, mean number of waiting grid jobs, expected sojourn time of jobs and so forth can be considered as performance measures [3,6,[8][9][10].…”