查看文章

Over the last few years, Grid technologies have progressed towards a service-oriented paradigm that enables a new way of service provisioning based on utility computing models. It provides services to users on the basis of pay-per-use. The main challenge in utility Grid is to satisfy the user QoS requirement, as well as minimizing the execution cost of application. In scientific and enterprise domains most user applications are constructed in form of workflow in which precedence relations between tasks are defined. Scheduling problem of workflow task on application is a challenging area. As it belongs to the class of NP-hard problems, therefore the use of heuristics is the defacto approach in order to cope with its difficulty. The number of existing scheduling algorithms are usually dedicated for following utility Grid enviroimient: 1) fully coimected network, and 2) fixed data transfer time and cost (independent fi" om selected services). These scheduling algorithms are not suitable for arbitrary service networks (type of utility Grid environment), where services are distributed arbitrary.

In our work, we propose the cost based workflow scheduling on arbitrary service network. In arbitrary service network the communication time and cost, depends on the available path and bandwidth between services. The proposed workflow scheduling algorithms work in three phases: 1) all-pair bandwidth analysis phase, in which findmg the available path and bandwidth between services, 2) deadline distribution phase, in which the sub-deadlines are assign to every tasks, and 3). Planning phase, in which overall deadline is distributed over individual tasks. Two different …