A cost and time balancing algorithm for scheduling parallel tasks on Computing Grid

Resource management and scheduling using economic-based principles and market-oriented models have been proven to be useful for scheduling tasks in Computing Grid. However, most studies only focus on the single objective of either cost or makespan. In this paper, we consider both cost and makespan, and focus on scheduling parallel tasks from Grid users considering a commodity market. A cost and time balancing algorithm, called MMCTB, which is based on Min-Min scheduling algorithm is proposed. The MMCTB minimizes and manages the execution cost and makespan of user tasks by a balancing formula and giving consideration to task urgency. The simulations reveal that the MMCTB has merits of needing less time and cost to complete the task, and high ratio of tasks executed within deadline by comparing with an enhanced Greedy heuristic.