Brokering service for supporting problem-oriented grid environment

This paper describes scheduling methods and algorithms of resources, and also the development on their base of the broker resource for search optimum resources in problem-oriented grid-environment. The developed scheduling algorithm considers additional knowledge about subject domain specifics of tasks and the representation about a workflow. The algorithm is based on a dominant sequence clustering algorithm (DSC). Unlike the original algorithm is that, for mapping tasks on the computing resources used by task graph coloring and the merging of tasks in a cluster is based on the availability of free slots on computing resources. Proposed a diphasic reservation method of resources and accounting problem of the problem parameters to estimate the time of its execution. Use cases of the broker resource are also given, process of resource allocation and architecture of the resource broker CAEBeans Broker are described.