Download now Free registration required
Resource performance in the Computational Grid is not only heterogeneous, but also changing dynamically. However, scheduling algorithms designed for traditional parallel and distributed systems, such as clusters, only consider the heterogeneity of the resources. In this paper, a workflow scheduling algorithm is proposed. This algorithm uses given QoS guidance to allocate tasks to distributed computer resources whose performance is subject to dynamic change and can be described by predefined probability distribution functions. This new algorithm works in an offline way that allows it to be easily set up and run with less cost. Simulations have been done to test its performance with different workflow and resource settings.
- Format: PDF
- Size: 271.3 KB