Predicting hardware usage in a computing system

• US 8,417,811 B1
• Filed: 12/17/2009
• Issued: 04/09/2013
• Est. Priority Date: 12/17/2009
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Jenkins, Jonathan A., Sehn, Timothy M.
• Primary Examiner(s): VU, VIET DUY
• Application Number:
  US12/640,331
• Time in Patent Office: 1,209 Days
• Field of Search: 709/223, 709/224, 709/226, 709/230, 709/250
• US Class Current: 709/224
• CPC Class Codes: G06F 11/3442 for planning or managing the needed capacity
• CPC Class Codes: G06F 11/3452 Performance evaluation by statistical analysis

Abstract

Disclosed are various embodiments of modeling hardware usage in a computing system. Usage factors affecting hardware usage of the computing system are identified. Usage factor data over a time period that corresponds to the usage factors is extracted. A regression analysis is performed on the historical usage factor data that determines a hardware usage function characterizing a relationship between the usage factors and historical hardware usage data.

Claims

1. A method for projecting hardware usage in a computing system, comprising the steps of:

extracting historical hardware usage data in a computing device from the computing system, the hardware usage data characterizing hardware usage of the computing system over an historical time period;

identifying a plurality of usage factors affecting hardware usage of the computing system, the plurality of usage factors comprising at least one rate factor corresponding to a rate at which operations in the computing system occur over a unit of time, and a cumulative factor corresponding to an amount of data storage consumed by the computing system;

extracting historical usage factor data corresponding to at least one usage factor in the computing device from the computing system, the usage factor data characterizing occurrences of the at least one usage factor over the historical time period; and

performing at least one regression analysis on the historical usage factor data in the computing device, the regression analysis determining a first hardware usage function characterizing a relationship between the at least one usage factor and the historical hardware usage data.

2. The method of claim 1, further comprising the step of projecting future hardware usage based at least upon the first hardware usage function.

3. The method of claim 1, wherein the step of performing at least one regression analysis further comprises:

performing a plurality of regression analyses in the computing device on a plurality of historical usage factor data, the plurality of historical usage factor data comprising a plurality of data sets corresponding to a plurality of permutations of usage factors;

identifying a first permutation of usage factors in the computing device yielding a hardware usage function having the lowest variance from the historical hardware usage data; and

associating the hardware usage function having the lowest variance from the historical hardware usage data with a first time period.

4. The method of claim 3, further comprising the steps of:

performing a plurality of regression analyses and identifying a second permutation of usage factors yielding a hardware usage function having the lowest variance from the historical hardware usage data in a second time period;

determining a second hardware usage function having the lowest variance from the historical hardware usage data; and

associating the second hardware usage function with the second time period.

5. The method of claim 1, wherein the content requests further comprise at least one of:

an order, a transaction, a page view, or network bandwidth consumption.

6. The method of claim 1, wherein the computing system further comprises an electronic commerce system and the total amount of data stored in the computing system further comprises at least one of:

products, orders, customers, product data, or customer data.

7. The method of claim 1, wherein the first hardware usage function further comprises:

HW_usage=α

*Cumulative+β

*Rate+γ

;

wherein HW_usage comprises a total hardware usage metric, α

comprises a first hardware usage metric per unit of the cumulative factor, β

comprises a second hardware usage metric per unit of the rate factor, and γ

comprises a baseline hardware usage metric if the cumulative factor and rate factor are zero.

8. The method of claim 1, wherein the computing system further comprises a system selected from the group consisting of a content delivery system, a search system and a data repository.

9. A system, comprising:

at least one computing device; and

a hardware usage monitor application executable in the at least one computing device, the hardware usage monitor application comprising;

logic that selects at least one rate factor and at least one cumulative factor affecting the hardware usage of a computing system, the at least one rate factor related to a rate at which content requests to the computing system occur over a time unit, and the cumulative factor related to an amount of data storage consumed by the computing system;

logic that performs at least one regression analysis on usage factor data regarding the at least one rate factor and at least one cumulative factor over a first time period, the regression analysis determining a relationship between the at least one rate factor, the at least one cumulative factor, and hardware usage data over the first time period; and

logic that generates at least one hardware usage function characterizing the relationship.

10. The system of claim 9, wherein the hardware usage monitor further comprises logic that projects future hardware usage based at least upon the first hardware usage function.

11. The system of claim 9, wherein the logic that selects at least one rate factor and at least one cumulative factor further comprises logic that selects a plurality of rate factors and a plurality of cumulative factors;

and

the logic that performs at least one regression analysis further comprises logic that performs a plurality of regression analyses on permutations of the rate factors and the cumulative factors.

12. The system of claim 11, wherein the logic that generates at least one hardware usage function further comprises logic that selects a first hardware usage function having the lowest variance relative to hardware usage data over the first time period.

13. The system of claim 12, wherein the logic that performs at least one regression analysis further comprises logic that performs a plurality of regression analyses on permutations of the rate factors and the cumulative factors over a second time period.

14. The system of claim 13, further comprising:

logic that selects a second hardware usage function having the lowest variance relative to hardware usage data over the second time period; and

logic that generates an alert if the first hardware usage function is different from the second hardware usage function.

15. The system of claim 9, wherein the hardware usage function further comprises:

HW_usage=α

*Cumulative+β

*Rate+γ

;

wherein HW_usage comprises a total hardware usage metric, α

comprises a first hardware usage metric per unit of the at least one cumulative factor, β

comprises a second hardware usage metric per unit of the at least one rate factor, and γ

comprises a baseline hardware usage metric if the at least one cumulative factor and at least one rate factor are zero.

16. The system of claim 9, wherein the computing system further comprises a system selected from the group consisting of a content delivery system, a search system and a data repository.

17. A computer readable medium including a hardware usage monitor application executable by at least one computing device, the hardware usage monitor application comprising:

logic that extracts historical hardware usage data from the computing system, the hardware usage data characterizing hardware usage of the computing system over an historical time period;

logic that identifies a plurality of usage factors affecting hardware usage of the computing system, the plurality of data storage consumed by the computing system;

logic that extracts historical usage factor data corresponding to at least one usage factor from the computing system, the usage factor data characterizing occurrences of the at least one usage factor over the historical time period;

logic that performs at least one regression analysis on the historical usage factor data, the regression analysis determining a hardware usage function characterizing a relationship between the at least one usage factor and the historical hardware usage data; and

logic that calculates a variance of the hardware usage function relative to a plot of the historical usage data over a time period.

18. The computer readable medium of claim 17, further comprising logic that projects future hardware usage based at least upon the first hardware usage function.