KReSIT Logo Kanwal Rekhi School of Information Technology
IIT Bombay
 Student Projects
 Ongoing Projects
 Past Projects
 Proposed Projects-2005
 Projects Archives
 ASL BE/MCA Projects
 Sponsored Projects

Home > Research > Projects > Students > Previous 

Students' Projects

Topology-Aware Failure Diagnosis For Distributed Enterprise Systems.

Amar Agrawal


The increasing complexity of modern enterprise systems has necessitated the automation of various management tasks. Such self-managing systems - also known as autonomic systems - can be self-healing, self-configuring, self-optimizing and self-securing. In this project we focus on the self-healing aspect, so that applications could detect and diagnose failures inorder to isolate the root-cause of these failures. Several approaches have been proposed for root-cause analysis including expert systems, data mining techniques, etc. However, none of these use the application topology information to help the failure diagnosis process. We believe that the application dependency information can be used to make the diagnosis process more focussed and accurate even in the presence of uncertainty (eg. lost/spurious symptoms). We model the system failure dependency information in terms of a Bayesian Belief Network and use it to infer the system state based on the observed failure symptoms in the application. The output of the project would be a tool that can monitor applications for failures and provide a component ranking based on inferred failure probabilities.

Printer friendly    Comment
  Copyright © 2004 KReSIT, IIT Bombay. All rights reserved sitemap    
  Kanwal Rekhi School of Information Technology, Indian Institute of Technology Bombay, Powai, Mumbai - 400 076.
+91-22-2576 7901/02. Fax: +91-22-2572 0022
Designed by Kamlesh