Computer system and method for service load distributing

• US 20050050198A1
• Filed: 08/26/2004
• Published: 03/03/2005
• Est. Priority Date: 08/29/2003
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Digital Solutions Corporation (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Mizoguchi, Kenichi
• Granted Patent: US 7,668,935 B2
• Time in Patent Office: Days
• US Class Current: 709/224
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system

Abstract

In a computer system executing a plurality of types of services, a service load monitor measures a load required to execute services in each of the computers. A node load monitor measures a load on each of the computers. A service optimal allocation machine determines an optimal computer and a service to be reallocated to the optimal computer, based on the measurement results of the service load monitor and the node load monitor, and reallocates the determined service to the optimal computer.

Claims

1. A computer system including a plurality of computers and executing a plurality of types of services, comprising:

service load monitor means for measuring, as a service load, a load required to execute services in each of the computers;

node load monitor means for measuring, as a node load, a load on each of the computers; and

service optimal allocation means for determining an optimal computer included in the computers to execute services, and a service to be reallocated to the optimal computer, based on measurement results of the service load monitor means and the node load monitor means, the service optimal allocation means reallocating the determined service to the determined optimal computer.

2. The computer system according to claim 1, wherein:

the service load monitor means includes means for calculating, as the service load, dynamic service ticket values based on an amount of resources used when each of the computers executes services, the dynamic service ticket values indicating dynamic loads of the services;

the node load monitor means includes means for calculating a total service ticket value indicating the node load of each of the computers, based on the dynamic service ticket values of the services executed in each of the computers, and means for calculating a dynamic node ticket value as a new ticket value usable in each of the computers, based on the total service ticket value of each of the computers and a static node ticket value indicating a preset processing capacity of each of the computers; and

the service optimal allocation means includes means for selecting a service to be reallocated to the optimal computer, from the services executed in at least one of the computers in which the dynamic node ticket value may become a value not more than a preset value.

3. The computer system according to claim 2, wherein the service optimal allocation means further includes means for searching for the optimal computer, to which the service selected by the selecting means is to be reallocated, based on the dynamic node ticket value of each of the computers.

4. The computer system according to claim 2, wherein the service optimal allocation means further includes means for searching for, as the optimal computer, one of the computers in which the dynamic node ticket value is higher than a given value, based on the dynamic node ticket value of each of the computers, and a higher one of a static service ticket value of the selected service and the dynamic service ticket value corresponding to the selected service executed in each of the computers, the given value being the higher one, the static service ticket value indicating a resource amount pre-estimated to be needed for executing the selected service in each of the computer.

5. The computer system according to claim 2, which further comprises parallel-execution-type service executing means for making at least two of the computers to execute a preset parallel-execution-type service in parallel,

and wherein the service optimal allocation means further includes:

means for calculating an optimal number of parallel executions of the parallel-execution-type service in the system, based on a dynamic service ticket value of the parallel-execution-type service executed by each of the computers, and a static service ticket value indicating a resource amount pre-estimated to be needed for executing the parallel-execution-type service in each of the computers, and

means for adjusting a current number of parallel executions of the parallel-execution-type service currently performed in the system by the parallel-execution-type service executing means, in accordance with a difference between the current number and the optimal number.

6. The computer system according to claim 5, wherein the adjusting means includes means for selecting, if the optimal number is larger than the current number, a computer to be controlled by the parallel-execution-type service executing means to execute the parallel-execution-type service, from those of the computers which can newly execute the parallel-execution-type service, in order beginning from a largest difference between the static node ticket value and the dynamic node ticket value.

7. The computer system according to claim 6, wherein the adjusting means includes means for selecting, if the optimal number is smaller than the current number, a computer in which execution of the parallel-execution-type service is to be stopped, from those of the computers in which execution of the parallel-execution-type service can be stopped, in order beginning from a smallest difference between the static node ticket value and the dynamic node ticket value.

8. The computer system according to claim 5, wherein the adjusting means includes means for selecting, if the optimal number is smaller than the current number, a computer in which execution of the parallel-execution-type service is to be stopped, from those of the computers in which execution of the parallel-execution-type service can be stopped, in order beginning from a smallest difference between the static node ticket value and the dynamic node ticket value.

9. A method of dynamically distributing services, for use in a computer system including a plurality of computers and executing a plurality of types of services, comprising:

acquiring service load information indicating loads of services executed in each of the computers;

acquiring node load information indicating a load on each of the computers;

determining an optimal computer included in the computers to execute services, and a service to be reallocated to the optimal computer; and

reallocating the determined service to the determined optimal computer.

10. The method according to claim 9, wherein:

the acquiring the service load information includes calculating, as the service load information, dynamic service ticket values based on an amount of resources used when each of the computers executes services, the dynamic service ticket values indicating dynamic loads of the services;

the acquiring the node load information includes calculating a total service ticket value indicating the node load information of each of the computers, based on the dynamic service ticket values of the services executed in each of the computers, and calculating a dynamic node ticket value as a new ticket value usable in each of the computers, based on the total service ticket value of each of the computers and a static node ticket value indicating a preset processing capacity of each of the computers; and

the determining includes selecting a service to be reallocated to the optimal computer, from the services executed in at least one of the computers in which the dynamic node ticket value may become a value not more than a preset value.

11. The method according to claim 10, wherein the determining further includes searching for, as the optimal computer, one of the computers in which the dynamic node ticket value is higher than a given value, based on the dynamic node ticket value of each of the computers, and a higher one of a static service ticket value of the selected service and one of the dynamic service ticket values corresponding to the selected service executed in each of the computers, the given value being the higher one, the static service ticket value indicating a resource amount pre-estimated to be needed for executing the selected service in each of the computer.

12. The method according to claim 10, further comprising:

executing, in parallel, a preset parallel-execution-type service using at least two of the computers;

calculating an optimal number of parallel executions of the parallel-execution-type service in the system, based on a dynamic service ticket value of the parallel-execution-type service executed by each of the computers, and a static service ticket value indicating a resource amount pre-estimated to be needed for executing the parallel-execution-type service in each of the computers; and

adjusting a current number of parallel executions of the parallel-execution-type service currently performed in the system, in accordance with a difference between the current number and the optimal number.

13. A computer program product for enabling dynamic load distributing of services in a computer system executing a plurality of types of services, the computer system including a plurality of computers having a service load monitor and a node load monitor, comprising:

computer readable program code means for acquiring, from the service load monitor, service load information indicating loads of services executed in each of the computers;

computer readable program code means for acquiring, from the node load monitor, node load information indicating a load on each of the computers;

computer readable program code means for determining an optimal computer included in the computers to execute services, and a service to be reallocated to the optimal computer; and

computer readable program code means for reallocating the determined service to the determined optimal computer.