Pattern Recognition in a Distributed Message Passing

Abstract

Recently, business intelligence has become a hot topic in corporation. It is a conceptual framework for decision supports. The goals for business intelli- gence projects are to provide information used to support internal enterprise users in the assessment, enhancement and optimization of organizational performance and operation and also to deliver critical business information to end-users. In this thesis, we analyze the idea of business intelligence in a distributed messaging system for utilization of current systems as well as providing information for end-users regarding their application behavior. We try to simulate the user application usage trends by help of queuing theory. Analyzing servers’ log files for extracting server resource usage pattern to decide if it is possible to use current servers without losing performance or shortage in resources. Another important functionality of the system is recognizing abnormal behavior of users’ applications in order to raise alarms or capacity planning to avoid system halts.