Load balancing for a parallel computer system by employing resource utilization target values and states

• US 5,898,870 A
• Filed: 12/13/1996
• Issued: 04/27/1999
• Est. Priority Date: 12/18/1995
• Status: Expired due to Fees

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Okuda, Tooru, Iwakura, Yoshiyuki, Nagasuka, Hirofumi
• Primary Examiner(s): Barry, Lance Leonard
• Application Number:
  US08/766,853
• Time in Patent Office: 865 Days
• Field of Search: 395/674, 395/675
• US Class Current: 718/104
• CPC Class Codes: G06F 9/505 considering the load

Abstract

A load sharing method for a parallel computer system having a computer group including a plurality of computers and an operation management mechanism which is a computer for managing the operation of the computer group. The method shares a load for executing a plurality of kinds of work processes to the plurality of computers in the computer group, and includes the steps of setting resource utilization target values by work for the plurality of computers in the computer group; collecting resource utilization states by work for the plurality of computers in the computer group to thereby inform the operation management mechanism of the resource utilization states; selecting a computer to execute a newly requested work process from the plurality of computers in the computer group on the basis of the differences between resource utilization target parameter values by work in the plurality of computers in the computer group and current values of a parameter indicating the reporting resource utilization states by work; and executing, in the selected computer, the newly requested work process

Claims

1. In a parallel computer system having a computer group including a plurality of computers and an operation management mechanism which includes a computer for managing the operation of the computer group, a load sharing control method for sharing a load for executing a plurality of kinds of work processes among said plurality of computers in said computer group, said method comprising the steps of:

setting a resource utilization target value by each work for each of said plurality of computers included in said computer group in said operation management mechanism;

collecting a parameter value indicative of a resource utilization state, corresponding to each work, for each of said plurality of computers included in said computer group into said operation management mechanism;

selecting one computer to execute a newly requested work process from said plurality of computers included in said computer group on the basis of the differences between resource utilization target parameter values by respective works in said plurality of computers included in said computer group and parameters values indicating current resource utilization states corresponding to the respective works from said operation management mechanism; and

executing, in the selected computer, the newly requested work process.

2. A load sharing control method according to claim 1, wherein said parameter value indicating the resource utilization state corresponding to each work of said plurality of computers included in said computer group collected by said operation management mechanism includes at least one of CPU utilization ratio information for each of said computers included in said computer group, CPU utilization ratio information weighted by difference in performance of said plurality of computers included in said computer group, and main storage available-state information in said plurality of computers included in said computer group.

3. A load sharing control method according to claim 1, wherein the step of selecting a computer to execute a newly requested work process includes a substep of calculating a difference between a resource utilization target parameter value corresponding to each work in each of said computers included in said computer group and the collected resource utilization state parameter corresponding to each work to thereby select a computer of which said difference is largest.

4. A load sharing control method according to claim 1, wherein said operation management mechanism includes a high-connection facility.

5. A load sharing control method according to claim 1, wherein said plurality of kinds of work processes includes at least two of an online process, a batch process and a time-sharing system process.

6. In a parallel computer system having a computer group including a plurality of computers and an operation management mechanism which is a computer for managing the operation of the computer group, a load sharing control apparatus for sharing a load for executing a plurality of kinds of work processes to said plurality of computers in said computer group comprises:

means provided in said operation management mechanism for setting a resource utilization target value corresponding to each work for each of said plurality of computers included in said computer group;

means provided in said operation management mechanism for collecting a parameter value indicative of resource utilization state corresponding to each work for each of said plurality of computers included in said computer group into said operation management mechanism;

means provided in each of said plurality of computers included in said computer group for selecting one computer to execute a newly requested work process from said plurality of computers included in said computer group on the basis of the differences between resource utilization target parameter values by respective works in said plurality of computers included in said computer group and current values of a parameter values indicating current resource utilization states by the respective works from said operation management mechanism; and

means for executing, in the selected computer, the newly requested work process.

7. A load sharing control apparatus according to claim 6, wherein said parameter value indicating the resource utilization state corresponding to each work of said plurality of computers included in said computer group collected by said operation management mechanism includes at least one of CPU utilization ratio information for each of said computers included in said computer group, CPU utilization ratio information weighted by difference in performance of said plurality of computers included in said computer group, and main storage available-state information in said plurality of computers included in said computer group.

8. A load sharing control apparatus according to claim 6, wherein said means for selecting a computer to execute a newly requested work process includes means for calculating a difference between a resource utilization target parameter value corresponding to each work in each of said computers included in said computer group and the informed resource utilization state parameter corresponding to each work to thereby select a computer of which said difference is largest.

9. A load sharing control apparatus according to claim 6, wherein said plurality of kinds of work processes includes at least two of an online process, a batch process and a time-sharing system process.

10. In a parallel computer system having a computer group including a plurality of computers and a shared storage unit for dynamically storing resource load information of the computer group which includes a computer for managing the operation of the computer group, a load sharing control method for sharing a load for executing a plurality of kinds of work processes among said plurality of computers in said computer group, said method comprising the steps of:

setting a resource utilization target value by each work for each of said plurality of computers included in said computer group in said operation management mechanism;

collecting a parameter value indicative of resource utilization states, corresponding to each work, for each of said plurality of computers included in said computer group into said operation management mechanism;

selecting one computer to execute a newly requested work process from said plurality of computers included in said computer group on the basis of the differences between resource utilization target parameter values by respective works in said plurality of computers included in said computer group and parameters values indicating current resource utilization states corresponding to the respective works from said operation management mechanism; and

executing, in the selected computer, the newly requested work process.

11. In a parallel computer system having a computer group including a plurality of computers and a shared storage unit for dynamically storing resource load information of the computer group which is a computer for managing the operation of the computer group, a load sharing control apparatus for sharing a load for executing a plurality of kinds of work processes to said plurality of computers in said computer group comprises:

means provided in said operation management mechanism for setting a resource utilization target value corresponding to each work for each of said plurality of computers included in said computer group;

means provided in said operation management mechanism for collecting a parameter value indicative of resource utilization state corresponding to each work for each of said plurality of computers included in said computer group into said operation management mechanism;

means provided in each of said plurality of computers included in said computer group for selecting one computer to execute a newly requested work process from said plurality of computers included in said computer group on the basis of the differences between resource utilization target parameter values by respective works in said plurality of computers included in said computer group and current values of a parameter values indicating current resource utilization states by the respective works from said operation management mechanism; and

means for executing, in the selected computer, the newly requested work process.