Abstract:

In a hybrid gridcloud environment,the user has elasticity provided by public cloud resources that can be aggregated to the existing grid resource pool as necessary.However,the integration opens new problem,such as:which are the best resources to loan from a public cloud based on current QoS requirement?Existing scheduling strategies based on static resource information cannot absolutely guarantee the successful completion of scientific workflow and satisfy the user’deadline requirements.In this paper,we propose an infrastructure,which is able to manage the execution of scientific workflow. Aimed at the scheduling of deadlinesensitive scientific workflow,the stochastic service model is adopt to model dynamic service capacity of grid resource and a metric called Default Risk of Task (DST) is offered to judge whether virtual resources should be loaned from cloud providers or not.Thus a scheduling algorithm (HCA_SASWD) is put forward.The experimental results show that HCA_SASWD achieves better performance than other algorithms on user’s deadline guarantee.

Key words: cloud;grid system;hybrid computing environment;workflow;deadline;default risk