• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.7 no.12
• /
• pp.3096-3117
• /
• 2013

To execute the performance driven Grid applications, an effective and scalable workflow scheduling is seen as an essential. To optimize cost & makespan, in this paper, we propose a Scalable Cost-Time Trade-off (SCTT) model for scheduling workflow tasks. We have developed a heuristic algorithm known as Scalable Cost-Time Trade-off Scheduling (SCTTS) with a lower runtime complexity based on the proposed SCTT model. We have compared the performance of our proposed approach with other heuristic and meta-heuristic based scheduling strategies using simulations. The results show that the proposed approach improves performance and scalability with different workflow sizes, task parallelism and heterogeneous resources. This method, therefore, outperforms other methods.

• Industrial Engineering and Management Systems
• /
• v.9 no.2
• /
• pp.70-79
• /
• 2010

Makespan and cost minimization are two important factors in project investment. This paper considers a multi-mode resource-constrained project scheduling problem with the objective of minimizing costs, subject to a deadline constraint. A number of studies have focused on minimizing makespan or resource availability cost with a specified deadline. This problem assumes a fixed cost for the availability of each renewable resource per period, and the project cost to be minimized is the sum of the variable cost associated with the execution mode of each activity. The presented memetic algorithm (MA) consists of three features: (1) a truncated branch and bound heuristic that serves as effective preprocessing in forming the initial population; (2) a strategy that maintains two populations, which respectively store deadline-feasible and infeasible solutions, enabling the MA to explore quality solutions in a broader resource-feasible space; (3) a repair-and-improvement local search scheme that refines each offspring and updates the two populations. The MA is tested via ProGen generated instances with problem sizes of 18, 20, and 30. The experimental results indicate that the MA performs exceptionally well in both effectiveness and efficiency using the optimal solutions or the current best solutions for the comparison standard.