Method for scheduling cloud-computing resource and system applying the same

1. A method for scheduling cloud-computing resource, comprising:

receiving a computing request at least including a number of the computing nodes, a start time of computing, and a computing length;

retrieving a computing resource table of each of a plurality of computing nodes, wherein the table buffered in a memory of a broker includes a computing pattern obtained from a load history record of each computing node, wherein every load history record is stored in every node's memory;

calculating an availability of each computing node according to the computing request and the computing pattern, wherein the availability indicates the available level of the each computing node, and the availability is obtained by summing up values of available resources while any load level of the computing pattern conforms with the computing length;

calculating a confidence of each computing node according to the computing request and the computing pattern, wherein the confidence shows the level of how the computing node matches up the computing request, and the confidence is obtained by:

calculating a first number of the computing pattern conforming with the computing length and the start time; calculating a second number of the computing pattern conforming with the start time and smaller than the computing length; and

the confidence equals to the first number divided by the second number;

calculating an expectation value of each computing node and the expectation value equals to the confidence multiplied by the availability; and

selecting one or more computing nodes according to the computing request and the expectation value of each computing node;

executing a computing task; and

distributing the computing task to the selected one or more computing nodes.

2. The method of claim 1 , wherein, based on the number of computing nodes, one or more computing nodes are selected to perform the computing task.

3. The method of claim 2 , wherein, further based on the expectation value of each computing node, the one or more computing nodes are selected.

4. The method of claim 1 , wherein the plurality of computing nodes form a group, through an agent program, the load history record of each computing node is recorded.

5. The method of claim 4 , wherein the broker retrieves the load history record recorded by the agent program over a network.

6. The method of claim 5 , wherein the broker includes a resource parameter table for recording a node ID of each computing node and the computing pattern, and the broker is used to obtain a number of the computing nodes, a start time of computing, and a computing length in accordance with the computing request, and the availability and the confidence based on the computing request.

7. The method of claim 5 , wherein the broker integrates results delivered from the plurality of computing nodes into a final result after finishing computations of the computing nodes.

8. The method of claim 1 , wherein the load history record includes load status of CPU resource and the corresponding time.

9. The method of claim 8 , wherein the load status of CPU resource by time is divided into a plurality of load levels and the load levels are arranged based on a chronological order in order to obtain the computing pattern.

10. The method of claim 1 , wherein, while the second number is calculated, the computing length is 1.

11. A system for scheduling cloud-computing resource, comprising:

a plurality of computing nodes, which is divided into one or more groups, and an agent program stored in memory of the each computing node of each group is executed to retrieve load information of each computing node and to build a computing resource table for each computing node;

a broker having memory recording the computing resource table of each computing node retrieved by the agent program, wherein the broker at least includes a node ID and a computing pattern and a database recording the load information retrieved from the agent program of the each computing node, and receives a computing request including a number of the computing nodes, a computing length, and a start time of computing, and the broker distributes computing task to the plurality of computing nodes according to the computing request;

wherein, the computing resource table records an availability, a confidence obtained according to the computing request, and an expectation value calculated based on the availability and the confidence, in which the availability indicates the available level of the each computing node, and the availability is obtained by summing up values of available resources while any load level of the computing pattern conforms with the computing length; the confidence shows the level of how the computing node matches up the computing request, and the confidence is obtained by:

calculating a first number of the computing pattern conforming with the computing length and the start time; calculating a second number of the computing pattern conforming with the start time and smaller than the computing length; and the confidence equals to the first number divided by the second number; and

the expectation value equals to the confidence multiplied by the availability;

wherein, the computing nodes and the broker are connected over a network.

12. The system of claim 11 , wherein the computing node is a terminal computer system.

13. The system of claim 11 , wherein the load history record of each computing node is recorded by time, and the computing pattern for each computing node is established based on the load history record.

14. The system of claim 11 , wherein the broker sorts the plurality of expectation values by size, and distributes computing task to the plurality of computing nodes over the network.