It is important to manage and allocate the resources effective in a distributed cluster environment. Static resource allocation strategy can not meet the requirements of resources and requests- dynamic changes. It can produce some resource fragmentation in the resource pool

resulting in low utilization of grid resources.Classification mining models and algorithms about dynamic resources allocation are proposed. There is a process in every resource management server

mining the classification rules in the cluster based on history data and executing resource allocation based on these rules. Experiments show that DRA algorithm can better adapt to the changes in the grid environment compared to other allocation strategies and algorithms

and it can increase resources utilization in dynamic environment.