Energy consumption optimization in a data-processing system

• US 8,776,075 B2
• Filed: 09/04/2011
• Issued: 07/08/2014
• Est. Priority Date: 10/29/2010
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Cosentino, Davide, MacLellan, Scot, Orlandi, Ivan, Trovato, Ignazio Flavio
• Primary Examiner(s): SWIFT, CHARLES M
• Application Number:
  US13/225,450
• Publication Number:
  US 20120110585A1
• Time in Patent Office: 1,038 Days
• Field of Search: None
• US Class Current: 718/104
• CPC Class Codes: G06F 1/206 comprising thermal management
• CPC Class Codes: G06F 9/4893 taking into account power or heat criteria power management in computers in general G06F1/3203; thermal management in computers in general G06F1/206
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A method for optimizing energy consumption in a data-processing system comprising a set of data-processing units is disclosed. In one embodiment, such a method includes indicating a set of data-processing jobs to be executed on a data-processing system during a production period. An ambient temperature expected for each data-processing unit during the production period is estimated. The method calculates an execution scheme for the data-processing jobs on the data-processing system. The execution scheme optimizes the energy consumed by the data-processing system to execute the data-processing jobs based on the ambient temperature of the data-processing units. The method then executes the data-processing jobs on the data processing system according to the execution scheme. A corresponding apparatus and computer program product are also disclosed.

Claims

1. A method for optimizing energy consumption in a data-processing system including a set of data-processing units, the method comprising:

indicating a set of data-processing jobs to be executed on the data-processing system during a production period;

estimating an ambient temperature expected for each data-processing unit during the production period, wherein estimating the ambient temperature comprises estimating a temperature trend for each data-processing unit during the production period;

calculating an execution scheme for the data-processing jobs on the data-processing system, the execution scheme optimizing energy consumed by the data-processing system to execute the data-processing jobs based on the ambient temperature of the data-processing units, wherein calculating the execution scheme comprises calculating, based on the temperature trend, an exploitation trend for exploiting a set of resources on each data-processing unit to optimize energy consumed by the data-processing units during the production period, and creating a plan of execution for executing the data-processing jobs on the data-processing units during the production period; and

executing the data-processing jobs on the data-processing system according to the execution scheme.

2. The method of claim 1, wherein the data-processing system includes at least one data-processing center at a corresponding location, and estimating an ambient temperature for each data-processing unit during the production period comprises estimating an external temperature expected at the location during the production period.

3. The method of claim 2, wherein estimating an external temperature expected at the location comprises retrieving a weather forecast for the location during the production period.

4. The method of claim 1, wherein the resources of each data-processing unit include a processing power of the data-processing unit.

5. The method of claim 4, wherein the processing power of each data-processing unit includes a processing frequency of the data-processing unit.

6. The method of claim 1, further comprising estimating a resource requirement of each data-processing unit by:

estimating a number of machine instructions of each data-processing job to be executed on the data-processing unit according to a nominal processing power of the data-processing unit and an expected duration of the job, the expected duration of the job being estimated according to an actual duration of previous executions of the job; and

summing the number of machine instructions of the data-processing jobs to be executed on the data-processing unit.

7. The method of claim 1, wherein calculating the exploitation trend of each data-processing unit comprises:

setting the exploitation trend to an opposite of the temperature trend during the execution window scaled by a scaling factor;

shifting the exploitation trend to have an absolute minimum thereof in the execution window equal to a null value; and

correcting the exploitation trend to have an integration thereof over the execution window equal to the resource requirement of the data-processing unit.

8. The method of claim 1, wherein each data-processing job is subject to a set of execution constraints, and calculating the exploitation trend of each data-processing unit comprises:

determining which data-processing units are unsuitable to execute the corresponding data-processing jobs in compliance with the execution constraints according to the exploitation trend of the data-processing unit; and

updating the exploitation trend of each unsuitable data-processing unit to make the data-processing unit suitable to execute the corresponding data-processing jobs in compliance with the execution constraints.

9. The method of claim 8, further comprising:

monitoring the execution of the data-processing jobs to detect which data-processing units are at risk of not executing the corresponding data-processing jobs in compliance with the execution constraints according to the exploitation trend; and

updating the exploitation trend of each at risk data-processing unit to make the data-processing unit capable of executing the corresponding data-processing jobs in compliance with the execution constraints.

10. The method of claim 9, wherein updating the exploitation trend of a data-processing unit that is one of unsuitable to execute the corresponding data-processing jobs and at risk of not executing the corresponding data-processing jobs comprises:

determining a further absolute minimum of the temperature trend in a further execution window of the corresponding data-processing jobs to be executed on the data-processing unit; and

increasing the exploitation trend of the data-processing unit in correspondence to the further absolute minimum.

11. The method of claim 1, wherein the data-processing system includes a plurality of data-processing units at different locations, and calculating an execution scheme for the data-processing jobs comprises selecting a data-processing unit to execute each data-processing job to optimize the energy consumed according to the ambient temperature of the plurality of data-processing units.

12. A computer program product for optimizing energy consumption in a data-processing system including a set of data-processing units, the computer program product comprising:

a non-transitory computer-readable storage medium storing computer instructions which when executed by at least one processor cause the at least one processor to;

indicate a set of data-processing jobs to be executed on the data-processing system during a production period;

estimate an ambient temperature expected for each data-processing unit during the production period, wherein estimating the ambient temperature comprises estimating a temperature trend for each data-processing unit during the production period;

calculate an execution scheme for the data-processing jobs on the data-processing system, the execution scheme optimizing energy consumed by the data-processing system to execute the data-processing jobs based on the ambient temperature of the data-processing units, wherein calculating the execution scheme comprises calculating, based on the temperature trend, an exploitation trend for exploiting a set of resources on each data-processing unit to optimize energy consumed by the data-processing units during the production period, and creating a plan of execution for executing the data-processing jobs on the data-processing units during the production period; and

execute the data-processing jobs on the data-processing system according to the execution scheme.

13. The computer program product of claim 12, wherein the data-processing system includes at least one data-processing center at a corresponding location, and estimating an ambient temperature for each data-processing unit during the production period comprises estimating an external temperature expected at the location during the production period.

14. The computer program product of claim 13, wherein estimating an external temperature expected at the location comprises retrieving a weather forecast for the location during the production period.

15. The computer program product of claim 12, wherein the resources of each data-processing unit include a processing power of the data-processing unit.

16. The computer program product of claim 15, wherein the processing power of each data-processing unit includes a processing frequency of the data-processing unit.

17. The computer program product of claim 12, wherein the computer instructions further cause the at least one processor to estimate a resource requirement of each data-processing unit by:

estimating a number of machine instructions of each data-processing job to be executed on the data-processing unit according to a nominal processing power of the data-processing unit and an expected duration of the job, the expected duration of the job being estimated according to an actual duration of previous executions of the job; and

summing the number of machine instructions of the data-processing jobs to be executed on the data-processing unit.

18. The computer program product of claim 12, wherein calculating the exploitation trend of each data-processing unit comprises:

setting the exploitation trend to an opposite of the temperature trend during the execution window scaled by a scaling factor;

shifting the exploitation trend to have an absolute minimum thereof in the execution window equal to a null value; and

correcting the exploitation trend to have an integration thereof over the execution window equal to the resource requirement of the data-processing unit.

19. The computer program product of claim 12, wherein each data-processing job is subject to a set of execution constraints, and calculating the exploitation trend of each data-processing unit comprises:

determining which data-processing units are unsuitable to execute the corresponding data-processing jobs in compliance with the execution constraints according to the exploitation trend of the data-processing unit; and

updating the exploitation trend of each unsuitable data-processing unit to make the data-processing unit suitable to execute the corresponding data-processing jobs in compliance with the execution constraints.

20. An apparatus for optimizing energy consumption in a data-processing system including a set of data-processing units, the apparatus comprising:

at least one processor; and

at least one memory device coupled to the at least one processor and storing computer instructions for execution on the at least one processor, the computer instructions causing the at least one processor to;

indicate a set of data-processing jobs to be executed on the data-processing system during a production period;

estimate an ambient temperature expected for each data-processing unit during the production period, wherein estimating the ambient temperature comprises estimating a temperature trend for each data-processing unit during the production period;

calculate an execution scheme for the data-processing jobs on the data-processing system, the execution scheme optimizing energy consumed by the data-processing system to execute the data-processing jobs based on the ambient temperature of the data-processing units, wherein calculating the execution scheme comprises calculating, based on the temperature trend, an exploitation trend for exploiting a set of resources on each data-processing unit to optimize energy consumed by the data-processing units during the production period, and creating a plan of execution for executing the data-processing jobs on the data-processing units during the production period; and

execute the data-processing jobs on the data-processing system according to the execution scheme.

21. The apparatus of claim 20, wherein the data-processing system includes at least one data-processing center at a corresponding location, and estimating an ambient temperature for each data-processing unit during the production period comprises estimating an external temperature expected at the location during the production period.