Failure prediction apparatus and method

• US 7,050,936 B2
• Filed: 11/30/2001
• Issued: 05/23/2006
• Est. Priority Date: 09/06/2001
• Status: Expired due to Term

• Current Assignee: Mavenir Ltd. (Mavenir Group Holdings Ltd.)
• Original Assignee: Comverse Limited (Mavenir Group Holdings Ltd.)
• Inventors: Levy, Dorron, Tibika, Pierre Gérard
• Primary Examiner(s): WACHSMAN, HAL D
• Application Number:
  US09/996,789
• Publication Number:
  US 20030046026A1
• Time in Patent Office: 1,635 Days
• Field of Search: 702181-188, 702/81, 702/179, 702/193, 702/56, 702/FOR.134, 702/FOR.135, 702/FOR.137, 702/FOR.139, 702/FOR.170, 702/FOR.171, 703/2, 703/6, 703/13, 703/22, 706903-908, 706/910, 706911-918, 706/922, 706/932, 706/25, 706/21, 700/51, 700/79, 700/80, 714/1, 714/25, 714/37, 714/38, 714/42, 714/46, 714/47, 717/124, 717/126, 717/127
• US Class Current: 702/181
• CPC Class Codes: G06F 11/008 Reliability or availability analysis

Abstract

An apparatus and method for predicting failure in a system is provided having a measurement unit for repeatedly measuring a disorder indicator of the system, and a comparator for comparing obtained measurements of the disorder indicator with a predetermined statistical description of the disorder indicator. The comparison is used to determine whether a deviation is present between presently measured values of the disorder indicator and the statistical description. The apparatus issues a failure prediction upon determination of the presence of such a deviation. The apparatus is able to use routine monitoring of the system and relies on its own statistical analysis of the behavior of the disorder indicator, to provide fault prediction which is non-specific to the type of fault.

Claims

1. Apparatus for predicting failure in a system, the apparatus comprising:

a measurement unit for repeatedly measuring a disorder indicator of said system, wherein the disorder indicator represents a non-designated output of said system, the non-designated output represents waste of said system, and

a comparator for comparing obtained measurements of said disorder indicator with a predetermined statistical description of said disorder indicator to determine whether a deviation is present between presently measured values of said disorder indicator and said predetermined statistical description, said apparatus being operable to issue a failure prediction upon determination that such a deviation is statistically significant,

wherein said disorder indicator provides non-specific failure indications.

2. Apparatus according to claim 1, wherein said measurement unit is operable to measure said disorder indicator via a communication link, thereby to monitor remotely located systems.

3. Apparatus according to claim 1, further comprising a communication unit for alerting a call center in the event of a failure prediction.

4. Apparatus according to claim 1, applicable to the system without regard to a level of complexity of said system.

5. The apparatus according to claim 1, wherein the apparatus is operable to determine whether to issue the failure prediction when the measured disorder indicator is not a normal distribution pattern.

6. The apparatus according to claim 1, wherein the apparatus is operable to determine whether to provide the non-specific failure indications when said measured disorder indicator behaves in a surge function pattern.

7. Apparatus for predicting failure in a system, the apparatus comprising:

a measurement unit for repeatedly measuring a disorder indicator of said system, wherein the disorder indicator represents a non-designated output of said system, and the non-designated output represents waste of said system,

a statistical unit for building up a statistical description of said disorder indicator using measurements taken via said measurement unit during a training phase of operation of said system, and

a system thresholder, for using said statistical description to apply thresholds to said disorder indicator to predict system failure,

wherein said disorder indicator provides failure indications of general abnormalities.

8. A method of failure prediction comprising:

repeatedly measuring a disorder indicator of a system, wherein the disorder indicator represents a non-designated output of said system, and the non-designated output represents waste of said system,

comparing said disorder indicator with a statistical description of idealized behavior of said disorder indicator,

determining from said comparison whether a deviation is present in behavior of said disorder indicator, and

issuing an alert in the event of determination of such a deviation being of statistical significance,

wherein said disorder indicator provides non-specific failure indications based on a top down approach.

9. Method according to claim 8, wherein said measuring is carried out remotely.

10. Method according to claim 8, further comprising alerting a call center in the event of a failure prediction.

11. Method according to claim 8, wherein the method is applicable to the system without regard to a level of complexity of said system.

12. A method of failure prediction in a software system, the method comprising:

selecting a measurable indicator of a level of disorder in said software system, wherein said measurable indicator provides non specific failure indication, and is selected without detailed knowledge of said software system,

obtaining a statistical description of behavior of said measurable indicator within said software system,

repeatedly measuring said measurable indicator during operation of said software system, wherein said measurable indicator represents a non-designated output of said software system,

comparing said measurable indicator with said statistical description,

determining from said comparison whether a deviation is present in behavior of said measurable indicator, and

issuing an alert in the event of determination of such a deviation being of statistical significance.

13. A data carrier having stored thereon computer executable instructions which when combined with a general purpose computer is operable to provide:

a measurement unit for repeatedly measuring a disorder indicator of a software system, wherein the disorder indicator represents a non-designated output of said software system and is a byproduct of operation of said software system, and

a comparator for comparing obtained measurements of said disorder indicator with a predetermined statistical description of said disorder indicator to determine whether a deviation is present between presently measured values of said disorder indicator and said predetermined statistical description, said combination being operable to issue a failure prediction upon determination that such a deviation is statistically significant.

14. Apparatus for measuring quality of software operating in a system, the apparatus comprising:

a measurement unit for repeatedly measuring a disorder indicator of said system, wherein the disorder indicator represents a non-designated output of said system, and the non-designated output represents waste of said system, and

a comparator for comparing obtained measurements of said disorder indicator with a predetermined statistical description of said disorder indicator to determine whether a deviation is present between presently measured values of said disorder indicator and said predetermined statistical description, said apparatus being operable to issue a quality score of said software based on an extent of said deviation,

wherein said disorder indicator gives statically viable but non-specific failure indications.

15. A method for predicting failure in a computing system comprising:

automatically determining a disorder indicator representing waste that said computing system produces;

repeatedly measuring the disorder indicator of said computing system;

comparing obtained measurements of said disorder indicator with a predetermined statistical description of said disorder indicator to determine whether a deviation is present between presently measured values of said disorder indicator and said predetermined statistical description, wherein said disorder indicator predicts faults that are a result of ill-defined or unexpected phenomena; and

issuing a failure prediction upon determination that such a deviation is statistically significant.

16. The method according to claim 15, wherein said disorder indicator is determined without any detailed knowledge of said computing system.

17. The method according to claim 16, wherein said disorder indicator is determined without any detailed knowledge of operation of said computing system.

18. The method according to claim 17, wherein said disorder indicator is determined without any detailed knowledge of failure modes of said computing system.

19. The method according to claim 15, wherein said automatically determining a disorder indicator comprises:

observing the computing system for patterns in a behavior of said computing system that are indicative of faults.

20. The method according to claim 15, wherein said computing system is a customized computing system.

21. The method according to claim 15, wherein measuring said disorder indicator comprises measuring a number of failure messages, and wherein the deviation is present when an overall number of messages statistically significantly increases.

22. The method according to claim 15, wherein measuring said disorder indicator comprises measuring a number of logged faults, and wherein the measured number of logged faults are compared with an average distribution of the logged faults.

23. The method according to claim 15, wherein measuring said disorder indicator comprises measuring heat produced by said computing system and an overall load of said computing system, wherein for said comparison said measurement of said produced heat is adjusted based on said measurement of said overall load of said computing system.

24. The method according to claim 15, wherein measuring said disorder indicator comprises measuring internal system load produced by a central processing unit and an external system load produced by a messaging signaling unit, wherein for said comparison said measurement of said produced internal system load is adjusted based on said measurement of said external load of said computing system.

25. The method according to claim 15, wherein measuring said disorder indicator comprises measuring an amount of time a system resource is devoted to a particular task.

26. The method according to claim 15, wherein said computing system is a local area network system, and wherein in predicting failures said local area network system uses said disorder indicator in a top down approach.