ADJUSTING CLOUD RESOURCE ALLOCATION

• US 20150350108A1
• Filed: 06/03/2014
• Published: 12/03/2015
• Est. Priority Date: 06/03/2014
• Status: Active Grant

• Current Assignee: Kyndryl Incorporated
• Original Assignee: International Business Machines Corporation
• Inventors: BAUGHMAN, AARON K., O'Connell, Brian M., Van Der Stockt, Stefan
• Granted Patent: US 9,882,836 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: H04L 47/823 Prediction of resource usage
• CPC Class Codes: H04L 47/83 based on usage prediction

Abstract

A method, system, and computer program product for adjusting cloud resource allocation using n-tier simulation are provided in the illustrative embodiments. In a multi-tiered simulation configuration, a combination of predictive models is executed such that each tier executes at least one predictive model to produce a corresponding set of predicted events. Each tier simulates a process that is consuming a computing resource. Using a subset of a selected set of predicted events outputted from a corresponding selected tier, a set of features is extracted. each feature in the set of features has an effect on an outcome of the simulated process. The set of features is used in a demand level prediction model to predict a threshold demand. Reaching the threshold demand in an actual utilization of the computing resource is indicative of a likelihood of an unforeseen rise in a demand for the computing resource after a period.

Claims

1. A method for adjusting resource allocation in a cloud computing environment, the method comprising:

executing, in a multi-tiered simulation configuration, a combination of predictive models such that each tier in the multi-tiered simulation configuration executes at least one predictive model to produce a corresponding set of predicted events, wherein each tier in the multi-tiered simulation configuration simulates a process that is consuming a computing resource in the cloud;

extracting, using a subset of a selected set of predicted events outputted from a corresponding selected tier in the multi-tiered simulation configuration, a set of features, each feature in the set of features having an effect on an outcome of the simulated process; and

using the set of features in a demand level prediction model to predict a threshold demand, wherein reaching the threshold demand in an actual utilization of the computing resource is indicative of a likelihood of an unforeseen rise in a demand for the computing resource after a period.

2. The method of claim 1, further comprising:

concluding, by evaluating the actual utilization of the computing resource relative to the demand threshold at a first time, that the unforeseen rise will occur after the period after the first time; and

causing, responsive to the concluding, an adjustment in an allocation of the computing resource to the process.

3. The method of claim 2, further comprising:

using the set of features in a demand level prediction model to output a confidence level corresponding to the threshold demand, wherein the confidence level is indicative of a probability that when the threshold demand is reached, the unforeseen rise in the demand will also occur; and

determining whether the confidence level at least equals a predetermined level, wherein the concluding and the causing are responsive to the confidence level at least equaling the predetermined level.

4. The method of claim 1, further comprising:

producing from an (m−

1)th tier in the multi-tiered simulation configuration an (m−

1)th set of predicted events; and

producing from an m-th tier in the multi-tiered simulation configuration an m-th set of predicted events, wherein a predicted event in the (m−

1)th set is predicted to occur after a first period in future, wherein a predicted event in the m-th set is predicted to occur after a second period in the future, the second period being longer than the first period.

5. The method of claim 4, further comprising:

configuring the multi-tiered simulated configuration such that a starting time of the predicted event in the m-th set has a greater uncertainty than a starting time of the predicted event in the (m−

1)th set.

6. The method of claim 4, wherein a subset of the (m−

1)th set of predicted events forms an input of the m-th tier.

7. The method of claim 1, further comprising:

executing in one tier in the multi-tiered simulation configuration a domain-dependent predictive model that is trained using state information of a process being simulated.

8. The method of claim 1, further comprising:

executing in one tier in the multi-tiered simulation configuration a domain-independent predictive model that is trained using state information of a process not being simulated.

9. The method of claim 1, further comprising:

configuring one tier in the multi-tiered simulation configuration as a recall-oriented tier to execute at least one recall-oriented predictive model.

10. The method of claim 1, further comprising:

configuring one tier in the multi-tiered simulation configuration as a precision-oriented tier to execute at least one prediction-oriented predictive model.

11. The method of claim 1, further comprising:

executing in one tier in the multi-tiered simulation configuration at least one recall-oriented prediction model and at least one precision-oriented predictive model in one of (i) a series, (ii) parallel, and (iii) a combination of series and parallel.

12. The method of claim 1, further comprising:

identifying a variable of a feature in the set of features, wherein the variable described a component of a predicted event in the subset of a selected set of predicted events.

13. The method of claim 1, further comprising:

summarizing the set of features to form a feature summary, wherein the using the set of features in the demand level prediction model comprises using the feature summary in the demand level prediction model.

14. The method of claim 1, further comprising:

executing a base simulation prior to a first tier in the multi-tiered simulation configuration, wherein the base simulation uses a known fact of the simulated process as input.

15. The method of claim 1, wherein the method is embodied in a computer program product comprising one or more computer-readable tangible storage devices and computer-readable program instructions which are stored on the one or more computer-readable tangible storage devices and executed by one or more processors.

16. The method of claim 1, wherein the method is embodied in a computer system comprising one or more processors, one or more computer-readable memories, one or more computer-readable tangible storage devices and program instructions which are stored on the one or more computer-readable tangible storage devices for execution by the one or more processors via the one or more memories and executed by the one or more processors.

17. A computer program product for adjusting resource allocation in a cloud computing environment, the computer program product comprising:

one or more computer-readable tangible storage devices;

program instructions, stored on at least one of the one or more storage devices, to execute, in a multi-tiered simulation configuration, a combination of predictive models such that each tier in the multi-tiered simulation configuration executes at least one predictive model to produce a corresponding set of predicted events, wherein each tier in the multi-tiered simulation configuration simulates a process that is consuming a computing resource in the cloud;

program instructions, stored on at least one of the one or more storage devices, to extract, using a subset of a selected set of predicted events outputted from a corresponding selected tier in the multi-tiered simulation configuration, a set of features, each feature in the set of features having an effect on an outcome of the simulated process; and

program instructions, stored on at least one of the one or more storage devices, to use the set of features in a demand level prediction model to predict a threshold demand, wherein reaching the threshold demand in an actual utilization of the computing resource is indicative of a likelihood of an unforeseen rise in a demand for the computing resource after a period.

18. The computer program product of claim 17, further comprising:

program instructions, stored on at least one of the one or more storage devices, to conclude, by evaluating the actual utilization of the computing resource relative to the demand threshold at a first time, that the unforeseen rise will occur after the period after the first time; and

program instructions, stored on at least one of the one or more storage devices, to cause, responsive to the concluding, an adjustment in an allocation of the computing resource to the process.

19. The computer program product of claim 18, further comprising:

program instructions, stored on at least one of the one or more storage devices, to use the set of features in a demand level prediction model to output a confidence level corresponding to the threshold demand, wherein the confidence level is indicative of a probability that when the threshold demand is reached, the unforeseen rise in the demand will also occur; and

program instructions, stored on at least one of the one or more storage devices, to determine whether the confidence level at least equals a predetermined level, wherein the concluding and the causing are responsive to the confidence level at least equaling the predetermined level.

20. A computer system for adjusting resource allocation in a cloud computing environment, the computer system comprising:

one or more processors, one or more computer-readable memories and one or more computer-readable tangible storage devices;

program instructions, stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, to execute, in a multi-tiered simulation configuration, a combination of predictive models such that each tier in the multi-tiered simulation configuration executes at least one predictive model to produce a corresponding set of predicted events, wherein each tier in the multi-tiered simulation configuration simulates a process that is consuming a computing resource in the cloud;

program instructions, stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, to extract, using a subset of a selected set of predicted events outputted from a corresponding selected tier in the multi-tiered simulation configuration, a set of features, each feature in the set of features having an effect on an outcome of the simulated process; and

program instructions, stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, to use the set of features in a demand level prediction model to predict a threshold demand, wherein reaching the threshold demand in an actual utilization of the computing resource is indicative of a likelihood of an unforeseen rise in a demand for the computing resource after a period.