A Scalable Multi-Agent Infrastructure for Remote Failure Detection in Web Applications

Abstract

In this paper we present GOA-Net, a multi-agent architecture for remote detection of application-level failures in web-sites. It has been stated in the literature that despite the effort on system-level monitoring tools the internet applications and web-sites still face some application-level faults that end up to be seen by the end-user. This tool tries to overcome this problem and makes use of multiagent technology and possibly later internet-grid computing. The monitoring of web-site failures can be done through different views: inside the network of the application, from the internet backbone and from the perspective of the end-user. The tool includes some data-correlation policies to reduce the occurrence of false-alarms and provide more accurate results.