GCP: A load-balancing tool in e-Science environment

In e-science environment researchers can submit complex tasks as scientific workflows. However, because the network may be blocked or even interrupted sometimes, it can hinder the researchers to submit tasks, as well as the overload on the server. Additionally, many tasks submitted by researchers may use the same resources, which may result in the overload of the specific resources. Meanwhile, some of tasks may be the same, so it is not necessary to waste the resources and spend extra time to execute the tasks again. Furthermore, it is difficult for the researchers to select the most appropriate resources from multitudinous resources in the e-science environment. Therefore, in this paper we developed the grid computing pool (GCP) to solve the problems. It contains client buffer pool (CBP) and server computing pool (SCP). CBP deployed on each client is used to save the tasks temporarily to relieve the server load. SCP deployed in grid platform is used to optimize the resource allocation and task execution according to the current condition of the environment. Finally it shows simulations to evaluate the availability and efficiency of the GCP.