Scheduling method and system for job task of file

The application discloses a scheduling method for a job task of a file, which comprises the following steps of: searching nodes where a plurality of data blocks of a file needed to carry out the job task are located; calculating loads produced by the file on the nodes where respective data blocks are located and operation loads of the nodes; comparing the operation loads of the nodes, and acquiring a node with the lightest load as a preset node; and scheduling the job task to be carried out on the preset node when the sum of the load produced by the file on the preset node and the operation load of the preset node is smaller than the threshold of a set load. When the method provided in the application is adopted, the system can distribute the job task to be carried out on the calculated preset node when the task is delivered in the file, and therefore the local property of data is met to the maximum extent, the increase of IO (Input/Output) expense due to data movement during the parallel execution of tasks is reduced, and the loads of the system are more balanced.