System and method for managing faults in a distributed system

• US 5,664,093 A
• Filed: 07/25/1996
• Issued: 09/02/1997
• Est. Priority Date: 12/27/1994
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Chang, Hsuan, Barnett, Bruce Gordon, Hartman, Michael James, Crapo, Andrew Walter, Vivier, Barbara Jean, Bloomer, John Joseph
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Decady, Albert
• Application Number:
  US08/686,443
• Time in Patent Office: 404 Days
• Field of Search: 395/50, 395/54, 395/183.01, 395/183.02, 395/183.07, 364/274, 364/274.3, 364/274.5
• US Class Current: 714/31
• CPC Class Codes: G06F 11/142 Reconfiguring to eliminate the error group management mechanisms in a peer-to-peer network H04L67/1044

Abstract

A system and method for managing faults in a distributed system. The fault management system includes a configuration manager that maintains configuration information of components used in the distributed system. A plurality of measurement agents obtain performance information from the components in the distributed system. A diagnostic system has a plurality of rules written according to the configuration information stored therein. The diagnostic system is coupled to the configuration manager and each of the plurality of measurement agents and identifies faults occurring in the distributed system and provides solutions for correcting the faults The diagnostic system receives the configuration information from the configuration manager and the performance information from the plurality of measurement agents and uses the configuration and performance information with the plurality of rules to identify faults and provide solutions for the faults.

Claims

1. A fault management system for use in a distributed system, comprising:

a configuration manager maintaining configuration information of components used in the distributed system, the configuration information comprising an object-oriented model describing relationships between the components, wherein the object-oriented model maintains a list of the components as objects and an understanding of how the objects are related;

a plurality of measurement agents obtaining performance information from the components in the distributed system; and

a diagnostic system coupled to the configuration manager and each of the plurality of measurement agents for identifying faults occurring in the distributed system and providing solutions for correcting the faults, the diagnostic system comprising a knowledge base having a plurality of rules for the components and an inference engine for applying the rules to the performance information, the diagnostic system receiving the configuration information from the configuration manager and the performance information from the plurality of measurement agents and using the configuration and performance information to identify faults and provide solutions for the faults, the diagnostic system identifying faults by querying the configuration manager for the object-oriented model of the components and using the model along with the plurality of rules in the knowledge base to identify the causes responsible for the fault and to provide a solutions for correcting the faults, the diagnostic system initiating the identification of faults at any location in the object-oriented model.

2. The fault management system according to claim 1, wherein, the object-oriented model comprises a static model and a dynamic model.

3. The fault management system according to claim 1, wherein the components comprise hardware components, software components, actions, tasks, and operation results.

4. A method for managing faults occurring in a distributed system with a fault management system comprising a configuration manager maintaining configuration information of components used in the distributed system, a plurality of measurement agents obtaining performance information from the components in the distributed system, and an diagnostic system coupled to the configuration manager and each of the plurality of measurement agents for identifying faults occurring in the distributed system and providing solutions for correcting the faults, the method comprising the steps of:

developing an object-oriented model describing relationships between the components, wherein the object-oriented model includes a list of the components as objects and an understanding of how the objects are related;

identifying the component where a fault is being reported;

querying the configuration manager to obtain the object-oriented model describing the relationship of the reported faulty component with other components in the distributed system;

determining from the object-oriented model which components may be responsible for the reported fault, the determination of faults being initiated at any location in the object-oriented model;

examining the components and applying rules within the diagnostic system to the relationship described in the object-oriented model to identify causes responsible for the fault; and

providing solutions for correcting the faults.

5. The method according to claim 4, wherein, the object-oriented model comprises a static model and a dynamic model.

6. The method according to claim 4, wherein the components comprise hardware components, software components, actions, tasks, and operation results.