FRAMEWORK FOR DISTRIBUTION OF COMPUTER WORKLOADS BASED ON REAL-TIME ENERGY COSTS

• US 20100058350A1
• Filed: 09/03/2008
• Published: 03/04/2010
• Est. Priority Date: 09/03/2008
• Status: Active Grant

• Current Assignee: Kyndryl Incorporated
• Original Assignee: International Business Machines Corporation
• Inventors: Doran, James R, Boss, Gregory J., Hamilton, Rick A. II, Sand, Anne R.
• Granted Patent: US 8,214,843 B2
• Time in Patent Office: Days
• US Class Current: 718/104
• CPC Class Codes: G06F 9/5094 where the allocation takes 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

Energy costs for conducting compute tasks at diverse data center sites are determined and are then used to route such tasks in a most efficient manner. A given compute task is first evaluated to predict potential energy consumption. The most favorable real-time energy costs for the task are determined at the various data center sites. The likely time period of the more favorable cost as well as the stability at the data center are additional factors. A workload dispatcher then forwards the selected compute task to the data center having the most favorable real-time energy costs. Among the criteria used to select the most favorable data center is a determination that the proposed center presently has the resources for the task. A stabilizer is utilized to balance the workload among the data centers. A computer implementation for performing the various steps of the cost determination and allocation is also described.

Claims

1. A system for dynamically shifting compute workloads among data center sites based on real-time energy costs, comprising:

a) an evaluator to determine potential energy consumption for a compute task;

b) criteria to determine which of the data center sites provides the most favorable real-time energy costs; and

c) a workload dispatcher to forward the compute task to a given data center site based upon the determination of the most favorable real-time energy costs.

2. The system according to claim 1 further including criteria for predefining available data centers having the resources for the compute task.

3. The system according to claim 1 further including a monitor for tracking power costs.

4. The system according to claim 1 including the determination of the optimum location for conducting the compute task.

5. The system according to claim 1 further including a stabilizer for balancing the workload among available data centers.

6. The system according to claim 1 wherein the workload dispatcher includes the evaluator.

7. A method for dynamically shifting compute workloads among data center sites based on real-time energy costs, comprising:

a) evaluating potential energy consumption for a compute task;

b) determining which of the data center sites provides the most favorable real-time energy costs; and

c) forwarding a compute task to a given data center site based upon the determination of the most favorable real-time energy costs.

8. The method according to claim 7 further including criteria for predefining available data centers having the resources for the compute task.

9. The method according to claim 7 further including the step of tracking power costs.

10. The method according to claim 7 including determining the optimum data center location for conducting the compute task.

11. The method according to claim 7 further including stabilizing the workload among data center sites.

12. The method according to claim 7 wherein the potential energy consumption for a compute task is evaluated by:

determining the cost per watt and the duration at that cost;

determining the length of time for the compute task;

determining the cost of running the task in each possible location;

selecting a location with the lowest cost; and

re-evaluating redistribution of the task as often as energy costs change at any of the locations.

13. A computer-readable medium containing instructions, when implemented on a computer, for shifting a compute task among data center sites based on real-time energy costs, by a method comprising:

a) evaluating potential energy consumption for a compute task;

b) determining which of the data centers provides the most favorable real-time energy costs; and

c) forwarding the compute task to a given data center site based upon the determination of the most favorable real-time energy costs.

14. The computer-readable medium containing instructions according to claim 13 including the step of predefining available computation resources.

15. The computer readable medium containing instructions according to claim 13 including the step of tracking power costs.

16. The computer readable medium containing instructions according to claim 13 further including the step of selecting the optimum data center site for conducting the compute task.

17. The computer readable medium containing instructions according to claim 13 including the step of stabilizing the workload among multiple data center sites.

18. A computer product including a medium on which a computer program is recorded, said computer program including:

a first set of instructions for evaluating potential energy consumption for at least one compute task;

a second set of instructions for determining which of multiple data centers provides the most favorable real-time energy costs; and

a third set of instructions for dispatching the compute task to a given data center site.

19. The computer product according to claim 18 wherein the computer program further includes a set of instructions for continuously monitoring and updating energy costs.

20. The computer product according to claim 18 wherein the computer program further includes instructions for stabilizing workload distribution among multiple data centers.

21. A method for deploying a data center site for running a compute workload, comprising:

a) evaluating potential energy consumption for a compute task;

b) determining which of the data centers provides the most favorable real-time energy costs; and

c) forwarding the compute task to a given data center site based upon the determination of the most favorable real-time energy costs.

22. The method according to claim 21 further including criteria for predefining available data centers having the resources for the compute task.

23. The method according to claim 21 including the step of determining the optimum location for conducting the compute task.

24. The method according to claim 21 further including the use of a stabilizer to balance the workload.