Historical data based workload allocation

• US 20060259621A1
• Filed: 05/16/2005
• Published: 11/16/2006
• Est. Priority Date: 05/16/2005
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Ranganathan, Parthasarathy, Moore, Justin, Farkas, Keith
• Granted Patent: US 7,644,148 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: G06F 1/206 comprising thermal management
• CPC Class Codes: G06F 1/3203 Power management, i.e. event-based initiation of a power-saving mode
• CPC Class Codes: G06F 2209/5019 Workload prediction
• CPC Class Codes: G06F 9/505 considering the load
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/125 involving control of end-device applications over a network
• CPC Class Codes: H04L 67/535 Tracking the activity of the user network monitoring arrangements H04L43/00; recording of computer activity G06F11/34
• CPC Class Codes: H05K 7/20836 Thermal management, e.g. server temperature control
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

In a method of allocating workload among servers based upon historical data, a requested workload profile is received and is compared with a plurality of historical workload profiles. A historical workload profile that is within a predefined range of the requested workload profile and that corresponds to a substantially minimized resource usage for power and cooling is selected. In addition, workload among the servers is allocated according to the selected historical workload profile to thereby substantially minimize resource usage for power and cooling.

Claims

1. A method of allocating workload among servers based upon historical data, said method comprising:

receiving a requested workload profile;

comparing the requested workload profile with a plurality of historical workload profiles;

selecting a historical workload profile that is within a predefined range of the requested workload profile and that corresponds to a substantially minimized resource usage for power and cooling; and

allocating workload among the servers according to the selected historical workload profile to thereby substantially minimize resource usage for power and cooling.

2. The method according to claim 1, wherein the power usage levels of each of the historical workload profiles corresponds to a cost function, and wherein the step of selecting a historical workload profile further comprises selecting a historical workload profile with the lowest cost function.

3. The method according to claim 1, wherein the step of selecting a historical workload profile further comprises selecting a historical workload profile that corresponds to a substantially minimized resource usage for power and cooling for the servers.

4. The method according to claim 1, further comprising:

measuring results of the workload allocation in response to the allocation of workload among the servers; and

determining whether the results of the workload allocation are within a predefined range of expected results.

5. The method according to claim 4, further comprising:

removing the selected historical workload profile from consideration in response to the results of the workload allocation falling outside of the predefined range of expected results;

identifying at least one another historical workload profile that is within a predefined range of the requested workload profile; and

selecting one of the at least one another historical workload profile; and

allocating workload among the servers according to the selected one of the at least one another historical workload profile.

6. The method according to claim 5, wherein the power usage levels of each of the historical workload profiles corresponds to a cost function, and wherein the step of selecting one of the at least one another historical workload profile further comprises selecting the historical workload profile with the lowest cost function.

7. The method according to claim 1, further comprising:

creating a repository containing the plurality of historical workload profiles; and

wherein the step of comparing the requested workload profile with a plurality of historical workload profiles further comprises comparing the requested workload profile with the plurality of historical workload profiles contained in the repository.

8. The method according to claim 7, wherein the step of creating a repository further comprises:

collecting a statistically significant amount of data; and

populating the repository with the statistically significant amount of data, wherein the statistically significant amount of data includes a relatively wide range of differing historical workload profiles.

9. The method according to claim 7, wherein the step of creating the repository further comprises:

collecting server utilization and thermal profile data periodically and over a relatively long period of time;

determining power usage levels based upon the thermal profile data; and

storing the collected server utilization and thermal profile data and the determined power usage levels to create the repository.

10. The method according to claim 7, wherein the step of creating the repository further comprises:

conducting a structured experiment that covers a relatively wide range of server utilization and thermal profiles to thereby define a relatively wide range of historical workload profiles;

determining power usage levels corresponding to the historical workload profiles based upon the structured experiment; and

storing the historical workload profiles to create the repository.

11. The method according to claim 1, further comprising:

creating a repository containing the plurality of historical workload profiles; and

wherein the step of comparing the requested workload profile with a plurality of historical workload profiles further comprises extrapolating workload profiles from existing historical workload profiles.

12. The method according to claim 1, further comprising:

determining a power usage level corresponding to the requested workload profile;

comparing the power usage level to a range of power usage levels corresponding to the historical workload profiles; and

randomly allocating the workload among the servers in response to the power usage level being outside of the range of power usage levels.

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

identifying the historical workload profiles that are within the range of power usage levels in response to the power usage level of the requested workload profile falling within the range of power usage levels; and

wherein the step of selecting a historical workload profile further comprises selecting a historical workload profile from the identified historical workload profiles.

14. A system for allocating workload among servers based upon historical data, said system comprising:

a database containing a repository of the historical data in a matrix of historical workload profiles;

a controller configured to search the repository to select a historical workload profile that is within a predefined range of a requested workload profile and that corresponds to a substantially minimized power usage level, said controller being further configured to allocate the workload associated with the requested workload profile according to the selected historical workload profile to substantially minimize resource usage for power and cooling in performing the workload.

15. The system according to claim 14, wherein the repository stored in the database comprises a table listing a relatively wide range of historical workload profiles and the power usage levels corresponding to each of the historical workload profiles.

16. The system according to claim 14, wherein the controller is further configured to collect a statistically significant amount of data corresponding to the historical workload profiles and to populate the repository with the statistically significant amount of data.

17. The system according to claim 14, wherein the controller is further configured to populate the repository with workload profiles extrapolated from existing historical workload profiles.

18. The system according to claim 14, wherein the controller is further configured to determine whether the allocated workload produced expected results and to re-allocate the workload if the allocated workload did not produce expected results.

19. The system according to claim 14, wherein the controller is further configured to select a historical workload profile that corresponds to a substantially minimized resource usage for power and cooling for the servers.

20. A system for allocating workload among servers, said system comprising:

means for storing a repository of historical workload profiles;

means for comparing a requested workload profile with the historical workload profiles;

means for selecting a historical workload profile; and

means for allocating workload among the servers based upon the historical workload profile selected by the means for selecting.

21. The system according to claim 20, further comprising:

means for creating the repository.

22. A computer readable storage medium on which is embedded one or more computer programs, said one or more computer programs implementing a method of allocating workload among servers based upon historical data, said one or more computer programs comprising a set of instructions for:

receiving a requested workload profile;

comparing the requested workload profile with a plurality of historical workload profiles;

selecting a historical workload profile that is within a predefined range of the requested workload profile and that corresponds to a substantially minimized power usage level in cooling the servers; and

allocating workload among the servers according to the selected historical workload profile to thereby substantially minimize resource usage for power and cooling for the servers.