| 摘要 |
A method of executing a program in a network environment that includes a source machine and a plurality of target machines, in which the source machine determines target machine eligibility according to configured constraints and selects a target machine from a list of target machines eligible to execute the program. The source machine transmits to the selected target machine an execution command for the program, together with the list of target machines, and a handle. The selected target machine launches the program and routes input and output requests to the source machine. The selected target machine monitors execution of the program. If performance constraints for the program are exceeded, or in response to a user initiated request, the selected target machine preempts execution of the program and roves the program to a selected next available target machine of the network. The selected next available target machine begins executing the preempted program at the point where the selected target machine preempted execution. The selected next available machine can rove the program to another machine of the network. Programs are executed to a network rather than to a machine.
|
