Predictive monitoring with wavelet analysis

• US 8,352,788 B2
• Filed: 04/06/2010
• Issued: 01/08/2013
• Est. Priority Date: 07/20/2009
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Borghetti, Stefano, Della Corte, Gianluca, Gianfagna, Leonida, Perrone, Antonio, Sgro, Antonio M
• Primary Examiner(s): PATEL, KAMINI B
• Application Number:
  US12/754,961
• Publication Number:
  US 20110016360A1
• Time in Patent Office: 1,008 Days
• Field of Search: 714/25
• US Class Current: 714/25
• CPC Class Codes: G06F 11/004 Error avoidance G06F11/07 and subgroups take precedence
• CPC Class Codes: G06F 11/0754 by exceeding limits

Abstract

A computer implemented method, computer program product and system for monitoring a plurality of hardware or software system resources for identifying hidden trends in the behavior of the system resources, includes: collecting metrics of at least one system resource indicative of the behavior of at least one system resource; for each of the at least one system resource, determining a spectrum representative of a time-based signal of the collected metrics; performing a wavelet transform on each of the at least one spectrum; and analyzing the result of the wavelet transform to identify possible linear trends in the behavior of the at least one system resource.

Claims

1. A computer implemented method for monitoring a plurality of hardware or software system resources for identifying hidden trends in the behavior of the system resources, the method comprising:

collecting metrics of at least one system resource indicative of the behavior of at least one system resource;

for each of the at least one system resource, determining a spectrum representative of a time-based signal of the collected metrics;

performing a wavelet transform on each of the at least one spectrum;

analyzing the result of the wavelet transform to identify possible linear trends in the behavior of the at least one system resource; and

assigning to each identified linear trend a severity value indicative of a criticality of the related behavior;

assigning to each identified linear trend the severity value indicative of the criticality of the related behavior comprises;

estimating, by means of extrapolation of the linear trend, a moment in time when a predetermined threshold of the related system resource will be reached.

2. The method of claim 1, wherein the determining the spectrum representative of the time-based signal of the collected metrics comprises:

interpolating the collected metrics for building a continuous graph.

3. The method of claim 1, further comprising:

for each of the at least one system resource, reading a critical threshold related to each metric.

4. The method of claim 3 wherein the analyzing the result of the wavelet

transform to identify the possible linear trends in the behavior of the at least one system resource comprises:

estimating, by extrapolation of the linear trends, the time required by the system resource metrics to reach related critical thresholds.

5. The method of claim 1, wherein the criticality of behaviors

comprises an indication of how fast a predetermined threshold of the related system resource will be reached.

6. The method of claim 1, wherein the severity value is proportional to an estimated speed of approach to the predetermined thresholds.

7. The method of claim 6, wherein the reading the critical threshold related to each metric comprises:

reading the critical threshold entered by a system monitoring administrator.

8. A computer program product for monitoring a plurality of hardware or software system resources for identifying hidden trends in the behavior of the system resources, the computer program product comprising:

a non-transitory computer readable storage medium having computer readable program code embodied therewith, the computer readable program code comprising;

computer readable program code configured to;

collect metrics of at least one system resource indicative of the behavior of at least one system resource;

for each of the at least one system resource, determine a spectrum representative of a time-based signal of the collected metrics;

perform a wavelet transform on each of the at least one spectrum;

analyze the result of the wavelet transform to identify possible linear trends in the behavior of the at least one system resource; and

assign to each identified linear trend a severity value indicative of a criticality of the related behavior;

assign to each identified linear trend the severity value indicative of the criticality of the related behavior is further configured to;

estimate, by means of extrapolation of the linear trend, a moment in time when a predetermined threshold of the related system resource will be reached.

9. The computer program product of claim 8, wherein the computer readable program code configured to determine the spectrum representative of the time-based signal of the collected metrics is further configured to:

interpolate the collected metrics for building a continuous graph.

10. The computer program product of claim 8, wherein the computer readable program code is further configured to:

for each of the at least one system resource, read a critical threshold related to each metric.

11. The computer program product of claim 10, wherein the computer readable program code configured to analyze the result of the wavelet transform to identify the possible linear trends in the behavior of the at least one system resource is further configured to:

estimate, by extrapolation of the linear trends, the time required by the system resource metrics to reach related critical thresholds.

12. The computer program product of claim 8, wherein the criticality of behaviors comprises an indication of how fast a predetermined threshold of the related system resource will be reached.

13. The computer program product of claim 8, wherein the severity value is proportional to an estimated speed of approach to the predetermined thresholds.

14. The computer program product of claim 13, wherein the computer readable program code configured to read the critical threshold related to each metric is further configured to:

read the critical threshold entered by a system monitoring administrator.

15. A system, comprising:

a plurality of system resources, comprising hardware or software system resources; and

a computer for monitoring the plurality of system resources, wherein the computer;

collects metrics of at least one system resource indicative of the behavior of at least one system resource;

for each of the at least one system resource, determines a spectrum representative of a time-based signal of the collected metrics;

performs a wavelet transform on each of the at least one spectrum;

analyzes the result of the wavelet transform to identify possible linear trends in the behavior of the at least one system resource; and

assigns to each identified linear trend a severity value indicative of a criticality of the related behavior;

assigns to each identified linear trend the severity value indicative of the criticality of the related behavior comprises;

estimates, by means of extrapolation of the linear trend, a moment in time when a predetermined threshold of the related system resource will be reached.

16. The system of claim 15, wherein the determines the spectrum representative of the time-based signal of the collected metrics comprises:

interpolates the collected metrics for building a continuous graph.

17. The system of claim 15, wherein the computer further:

for each of the at least one system resource, reads a critical threshold related to each metric.

18. The system of claim 17, wherein the analyzes the result of the wavelet transform to identify the possible linear trends in the behavior of the at least one system resource comprises:

estimates, by extrapolation of the linear trends, the time required by the system resource metrics to reach related critical thresholds.

19. The system of claim 15, wherein the criticality of behaviors comprises an indication of how fast a predetermined threshold of the related system resource will be reached.