CAPACITY CONTROL

• US 20110302301A1
• Filed: 10/28/2009
• Published: 12/08/2011
• Est. Priority Date: 10/31/2008
• Status: Active Grant

• Current Assignee: HSBC Group Management Services Ltd. (HSBC Holdings plc)
• Original Assignee: HSBC Holdings plc
• Inventors: Lowes, Raymond, Davies, Jonathan
• Granted Patent: US 9,176,789 B2
• Time in Patent Office: Days
• US Class Current: 709/224
• CPC Class Codes: G06F 2209/5019 Workload prediction
• CPC Class Codes: G06F 2209/508 Monitor
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system

Abstract

There are described a system, apparatus and/or method according to which collected history data from monitorable and/or manageable entities within a server estate or data centre is used to tailor a model to the past behaviour and/or loading upon individual ones or parts of ones of the entities. The tailored model can then be used to predict a future value for one or more metrics included in the history data for an entity or part of an entity under consideration. The predicted value can then be used to determine whether some form of corrective action is required in respect of the entity, entity part or some other entity or entity part. Alternatively or in addition, predicted values for more than one entity or entity part can be aggregated to determine whether some form of corrective action is required in respect of the entity, entity part or some other entity or entity part. A corrective action can be applied on an automated or on a manual basis.

Claims

1. A computer implemented method for controlling a resource implemented by a computer system, the method comprising:

collecting operation data from each of a plurality of network resources, wherein different ones of the network resources are operable under different management systems, and wherein the operation data is independent of the management system of the network resource to which the operation data relates;

predicting, using the operation data, by the computer system a future loading on at least one of the network resources, wherein the predicting includes using operation data collected over time to provide an operation data history;

determining by the computer system, using the predicted future loading, whether a corrective action is required;

determining by the computer system, when the corrective action is required, a predetermined corrective action for at least one of the plurality of network resources; and

applying the predetermined corrective action by the computer system to the network resource and controlling the network resource responsive thereto.

2. The method of claim 1, wherein the network resources include a plurality of computer processing resources and wherein different ones of the computer processing resources are operable under different computer file management systems.

3. The method of claim 1, wherein different ones of the plurality of network resources operate using different operating systems, and wherein the operation data is independent of the operating system of the network resource to which the operation data relates.

4. The method of claim 1, wherein the corrective action includes at least one action selected from the group comprising:

archival of data, compression of data, deletion of data, provision of additional resource, suppression of a service, and allocation of underused active resource to backup/idle.

5. The method of claim 4, wherein provision of additional resource comprises one or more of reallocation of resource from one service to a different service, and activation of at least one of a backup and idle resource.

6. The method of claim 4, wherein suppression of a service comprises selecting a lower priority service and suppressing availability of that service to enable resource to be reallocated to a higher priority service.

7. The method of claim 1, wherein the predicting uses a trending process to identify resource usage trends.

8. The method of claim 7, wherein the trending process uses a linear regression or time series analysis to identify resource usage trends.

9. The method of claim 8, wherein the time series analysis comprises double exponential smoothing or triple exponential smoothing.

10. The method of claim 1, wherein the predicting comprises using a mean squared difference to indicate an accuracy measure for the predicted future loading.

11. The method of claim 1, wherein the predicting uses model descriptors derived from a previous iteration, adjusts the model descriptors based upon the latest data in the collection window and stores the updated model descriptors for use in a future iteration.

12. The method of claim 1, further comprising making at least one of the operation data history and the predicted future loading available to an administrator.

13. The method of claim 1, wherein the operation data history includes one data point for each measured metric to represent a given time period.

14. The method of claim 13, wherein the data point is selected from a plurality of values for the metric collected during the time period, or is an average value of a plurality of values for the metric during the time period.

15. The method of claim 1, wherein the operation data history includes data collected over a period of between 1 and 4 months.

16. The method of claim 1, wherein the determining also uses an accuracy measure for the predicted future loading to determine whether a corrective action can be applied based upon the predicted future loading.

17. A computer architecture operable to control a resource, the computer system comprising:

at least one computer database storing historical data and operable to store an operation data history comprising operation data collected from each of a plurality of network resources, wherein different ones of the network resources are operable under different management systems, and wherein the operation data is independent of the management system of the network resource to which the operation data relates;

a computer system in communication with the at least one computer database, wherein the computer system comprises the following components executable by the computer system;

a predictor operable to predict, using the operation data in the operation data history, a future loading on at least one of the network resources;

a determiner operable to determine, using the predicted future loading, whether a corrective action is required, and to determine, when the corrective action is required, a predetermined corrective action for at least one of the plurality of network resources; and

an instruction generator operable to generate an instruction to cause the predetermined corrective action to be applied to the network resource and controlling the network resource responsive thereto.

18. A computer program comprising computer implementable instructions to cause a programmable computer to become configured to execute instructions encoded in a non-transitory computer-readable memory device, the instructions comprising:

predicting, using the operation data, by the computer system a future loading on at least one of the network resources, wherein the predicting includes using operation data collected over time to provide an operation data history;

determining by the computer system, using the predicted future loading, whether a corrective action is required;

determining by the computer system, when the corrective action is required, a predetermined corrective action for at least one of the plurality of network resources; and

applying the predetermined corrective action by the computer system to the network resource and controlling the network resource responsive thereto.