Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Automated known problem diagnosis with event traces

Automated known problem diagnosis with event traces ciation between low-level system behavior and high-level problem and the trace-based problem diagnosis is a promising troubleshooting method with good accuracy and generality. This paper is organized as follows. In Section 2 we describe the design of the automated diagnosis system. We introduce the event tracing component in Section 3 and describe the classifier we use to learn from system behavior and predict root cause in Section 4. In Section 5 we report our observations on system event traces which help us to design noise filtering and canonicalization rules. In Section 6 we evaluate this approach with four problems having diverse root causes. In Section 7 we discuss some questions about the method. We introduce related work in Section 8. (3) The root cause and the corresponding solution will be sent to the troubleshooter, which can present a report of repair instructions to the user or even fix the problem automatically. Step (4) is responsible for learning the classifier for root cause identification. With a database of known problems and their root causes (from the past accumulation of diagnosis knowledge), a number of event traces can be collected for them offline and labeled with the corresponding root causes. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Automated known problem diagnosis with event traces

Association for Computing Machinery — Apr 18, 2006

Loading next page...
 
/lp/association-for-computing-machinery/automated-known-problem-diagnosis-with-event-traces-W6H00AYHKC

References (30)

Datasource
Association for Computing Machinery
Copyright
Copyright © 2006 by ACM Inc.
ISBN
1-59593-322-0
doi
10.1145/1217935.1217972
Publisher site
See Article on Publisher Site

Abstract

ciation between low-level system behavior and high-level problem and the trace-based problem diagnosis is a promising troubleshooting method with good accuracy and generality. This paper is organized as follows. In Section 2 we describe the design of the automated diagnosis system. We introduce the event tracing component in Section 3 and describe the classifier we use to learn from system behavior and predict root cause in Section 4. In Section 5 we report our observations on system event traces which help us to design noise filtering and canonicalization rules. In Section 6 we evaluate this approach with four problems having diverse root causes. In Section 7 we discuss some questions about the method. We introduce related work in Section 8. (3) The root cause and the corresponding solution will be sent to the troubleshooter, which can present a report of repair instructions to the user or even fix the problem automatically. Step (4) is responsible for learning the classifier for root cause identification. With a database of known problems and their root causes (from the past accumulation of diagnosis knowledge), a number of event traces can be collected for them offline and labeled with the corresponding root causes.

There are no references for this article.