System component failure diagnosis

• US 9,916,194 B2
• Filed: 10/01/2015
• Issued: 03/13/2018
• Est. Priority Date: 10/01/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Pang, Yan
• Primary Examiner(s): Bonzo, Bryce
• Assistant Examiner(s): GIBSON, JONATHAN D
• Application Number:
  US14/872,276
• Publication Number:
  US 20170097860A1
• Time in Patent Office: 894 Days
• CPC Class Codes: G06F 11/0709 in a distributed system consisting of a plurality of standalone computer nodes, e.g. clusters, client-server systems
• CPC Class Codes: G06F 11/0787 Storage of error reports, e.g. persistent data storage, storage using memory protection
• CPC Class Codes: G06F 11/079 Root cause analysis, i.e. error or fault diagnosis in a hardware test environment G06F11/22; in a software test environment G06F11/36
• CPC Class Codes: G06F 11/34 Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: G06N 5/04 Inference or reasoning models

Abstract

System component failure diagnosis is provided. In response to determining that a number of component failure cases corresponding to a complex system stored in a historical diagnosis database is less than a pre-defined threshold number of component failure cases, a system component failure root cause is predicted using a rule-based reasoning method. In response to determining that the number of component failure cases corresponding to the complex system is greater than or equal to the pre-defined threshold number of component failure cases, the system component failure root cause is predicted using the rule-based reasoning method and a machine learning method. The diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning method and the machine learning method is calculated using a cross-validation method. The performance of the rule-based reasoning method and the machine learning method is evaluated using a statistical process control method.

Claims

1. A computer-implemented method for client system component failure diagnosis, the computer-implemented method comprising:

determining, by a server computer, whether a number of component failure cases corresponding to a client data processing system stored in a historical diagnosis database is less than a pre-defined threshold number of component failure cases needed to run a machine learning diagnosis component of the server computer to predict a system component failure root cause within the client data processing system;

responsive to the server computer determining that the number of component failure cases corresponding to the client data processing system stored in the historical diagnosis database is less than the pre-defined threshold number of component failure cases needed to run the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system, running, by the server computer, a rule-based reasoning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system;

calculating, by the server computer, a diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component of the server computer based on historical data corresponding to the client data processing system stored in the historical diagnosis database;

verifying, by the server computer, the system component failure root cause within the client data processing system based on the diagnosis accuracy; and

fixing, by the server computer, the verified system component failure root cause within the client data processing system increasing performance of the client data processing system.

2. The computer-implemented method of claim 1 further comprising:

responsive to the server computer determining that the number of component failure cases corresponding to the client data processing system stored in the historical diagnosis database is greater than or equal to the pre-defined threshold number of component failure cases needed to run the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system, running, by the server computer, the rule-based reasoning diagnosis component and the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system; and

calculating, by the server computer, the diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component and predicted by the machine learning diagnosis component of the server computer using a cross-validation diagnosis accuracy estimation method.

3. The computer-implemented method of claim 2 further comprising:

determining, by the server computer, whether the machine learning diagnosis component of the server computer has a higher calculated diagnosis accuracy as compared to the rule-based reasoning diagnosis component of the server computer using a statistical process control method.

4. The computer-implemented method of claim 3 further comprising:

responsive to the server computer determining that the machine learning diagnosis component of the server computer does not have the higher calculated diagnosis accuracy as compared to the rule-based reasoning diagnosis component of the server computer, displaying, by the server computer, the diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component in a user interface; and

responsive to the server computer determining that the machine learning diagnosis component of the server computer does have the higher calculated diagnosis accuracy as compared to the rule-based reasoning diagnosis component of the server computer, displaying, by the server computer, the diagnosis accuracy of the system component failure root cause predicted by the machine learning diagnosis component in the user interface.

5. The computer-implemented method of claim 1 further comprising:

storing, by the server computer, a component failure case corresponding to the system component failure root cause of the client data processing system in the historical diagnosis database; and

responsive to the server computer determining that information in the component failure case corresponding to the system component failure root cause of the client data processing system needs review by a subject matter expert, sending, by the server computer, a notification to the subject matter expert to review the information in the component failure case corresponding to the system component failure root cause of the client data processing system stored in the historical diagnosis database.

6. The computer-implemented method of claim 1, wherein the server computer predicts the system component failure root cause by selecting a system component having a highest calculated confidence level.

7. The computer-implemented method of claim 1, wherein the server computer combines a similarity confidence level value, a frequency confidence level value, a timing confidence level value, and a component-function causal relationship confidence level value in the machine learning diagnosis component of the server computer to calculate an overall confidence level value corresponding to a system component that is most likely the system component failure root cause within the client data processing system.

8. The computer-implemented method of claim 7, wherein the server computer multiplies each of the similarity confidence level value, the frequency confidence level value, the timing confidence level value, and the component-function causal relationship confidence level value by a different weight to calculate the overall confidence level value corresponding to the system component that is most likely the system component failure root cause within the client data processing system, and wherein the server computer calculates each different weight under a constraint that the overall confidence level value is between zero and one.

9. The computer-implemented method of claim 1, wherein the pre-defined threshold number of component failure cases needed to run the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system is based on subject matter expert experience or computed by a heuristic method.

10. A server computer for client system component failure diagnosis, the server computer comprising:

a bus system;

a storage device connected to the bus system, wherein the storage device stores program instructions; and

a processor connected to the bus system, wherein the processor executes the program instructions to;

determine whether a number of component failure cases corresponding to a client data processing system stored in a historical diagnosis database is less than a pre-defined threshold number of component failure cases needed to run a machine learning diagnosis component of the server computer to predict a system component failure root cause within the client data processing system;

run a rule-based reasoning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system in response to determining that the number of component failure cases corresponding to the client data processing system stored in the historical diagnosis database is less than the pre-defined threshold number of component failure cases needed to run the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system;

calculate a diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component of the server computer based on historical data corresponding to the client data processing system stored in the historical diagnosis database;

verify the system component failure root cause within the client data processing system based on the diagnosis accuracy; and

fix the verified system component failure root cause within the client data processing system increasing performance of the client data processing system.

11. The server computer of claim 10, wherein the processor further executes the program instructions to:

run the rule-based reasoning diagnosis component and the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system in response to determining that the number of component failure cases corresponding to the client data processing system stored in the historical diagnosis database is greater than or equal to the pre-defined threshold number of component failure cases needed to run the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system; and

calculate the diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component and predicted by the machine learning diagnosis component of the server computer using a cross-validation diagnosis accuracy estimation method.

12. The server computer of claim 11, wherein the processor further executes the program instructions to:

determine whether the machine learning diagnosis component of the server computer has a higher calculated diagnosis accuracy as compared to the rule-based reasoning diagnosis component of the server computer using a statistical process control method.

13. A computer program product for system component failure diagnosis, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a server computer to cause the server computer to perform a method comprising:

determining, by the server computer, whether a number of component failure cases corresponding to a client data processing system stored in a historical diagnosis database is less than a pre-defined threshold number of component failure cases needed to run a machine learning diagnosis component of the server computer to predict a system component failure root cause within the client data processing system;

responsive to a the server computer determining that the number of component failure cases corresponding to the client data processing system stored in the historical diagnosis database is less than the pre-defined threshold number of component failure cases needed to run the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system, running, by the server computer, a rule-based reasoning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system;

calculating, by the server computer, a diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component of the server computer based on historical data corresponding to the client data processing system stored in the historical diagnosis database;

verifying, by the server computer, the system component failure root cause within the client data processing system based on the diagnosis accuracy; and

fixing, by the server computer, the verified system component failure root cause within the client data processing system increasing performance of the client data processing system.

14. The computer program product of claim 13 further comprising:

responsive to the server computer determining that the number of component failure cases corresponding to the client data processing system stored in the historical diagnosis database is greater than or equal to the pre-defined threshold number of component failure cases needed to run the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system, running, by the server computer, the rule-based reasoning diagnosis component and the machine learning diagnosis component of the server computer to predict the system component failure root cause within the client data processing system; and

calculating, by the server computer, the diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component and predicted by the machine learning diagnosis component of the server computer using a cross-validation diagnosis accuracy estimation method.

15. The computer program product of claim 14 further comprising:

determining, by the server computer, whether the machine learning diagnosis component of the server computer has a higher calculated diagnosis accuracy as compared to the rule-based reasoning diagnosis component of the server computer using a statistical process control method.

16. The computer program product of claim 15 further comprising:

responsive to the server computer determining that the machine learning diagnosis component of the server computer does not have the higher calculated diagnosis accuracy as compared to the rule-based reasoning diagnosis component of the server computer, displaying, by the server computer, the diagnosis accuracy of the system component failure root cause predicted by the rule-based reasoning diagnosis component in a user interface; and

responsive to the server computer determining that the machine learning diagnosis component of the server computer does have the higher calculated diagnosis accuracy as compared to the rule-based reasoning diagnosis component of the server computer, displaying, by the server computer, the diagnosis accuracy of the system component failure root cause predicted by the machine learning diagnosis component in the user interface.