Failure recognition, notification, and prevention for learning and self-healing capabilities in a monitored system

• US 7,823,029 B2
• Filed: 09/07/2005
• Issued: 10/26/2010
• Est. Priority Date: 09/07/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Labadie, Eric, Kobylinski, Krzysztof R.
• Primary Examiner(s): Baderman; Scott T
• Assistant Examiner(s): KO, CHAE M
• Application Number:
  US11/220,963
• Publication Number:
  US 20070055915A1
• Time in Patent Office: 1,875 Days
• Field of Search: 714/48, 714/38, 714/47, 714/23, 714/26
• US Class Current: 714/47.1
• CPC Class Codes: G06F 11/004 Error avoidance G06F11/07 and subgroups take precedence

Abstract

The present invention provides failure recognition, notification, and prevention for learning and self-healing capabilities in a monitored system. A system to collect monitoring data is monitored. A failure of the system is detected; A failure point for the detected failure in a data space defined by the monitoring data is identified and at least one predefined action with the identified failure point is associated. This process is repeated for a plurality of system failures. When a state of the system is determined to be approaching an identified failure point, the at least one predefined action associated with that identified failure point is performed.

Claims

1. A method for failure recognition, comprising:

monitoring a system to collect monitoring data;

detecting a failure of the system, wherein the failure includes an abnormal system state;

identifying a failure point for the detected failure in a data space defined by the monitoring data;

associating at least one predefined action with the identified failure point;

identifying values of the monitoring data in a neighborhood of the identified failure point;

dividing the neighborhood of the identified failure point into pre-failure data and post-failure data, wherein the pre-failure data and post-failure data identify pre-failure and post-failure states of the system, respectively;

dividing the pre-failure data into a plurality of ranges and associating a unique predefined action with each of said ranges;

determining if a state of the system reaches a point within a space defined by one of the plurality of ranges;

in response to said state of the system reaching a point within a space defined by one of the plurality of ranges, performing said unique predefined action associated with said one of the plurality of ranges;

wherein each of the unique predefined actions comprises an automatic self-healing preventive action and at least one of the unique predefined actions further comprises a notification, wherein the notification comprises an indicator of system unavailability and a period of unavailability; and

outputting a representation of the performed predefined action.

2. The method of claim 1, further comprising:

automatically avoiding the identified failure point during operation of the system.

3. The method of claim 1, further comprising:

detecting at least one additional failure of the system; and

repeating the detecting, identifying, and associating steps for each additional detected failure, wherein a plurality of identified failure points are identified, and wherein a unique predefined action is associated with each of the plurality of identified failure points.

4. The method of claim 3, further comprising:

determining if a state of the system is approaching any of the plurality of identified failure points.

5. The method of claim 4, further comprising:

if the state of the system is approaching any of the identified failure points, performing the predefined action associated with that failure point.

6. The method of claim 3, further comprising:

automatically avoiding any of the plurality of identified failure points during operation of the system.

7. The method of claim 1, wherein the data space defined by the monitoring data comprises a multi-dimensional data space.