Policy Driven Software Monitoring

Abstract

Software monitoring and logging is one of the most important tools a software engineer has when faced with the challenge of auditing or analysing a software system. However, the difficulty in effectively monitoring a system, managing its logs and cross referencing them with source code makes software re-engineering a rigorous and complex task. This thesis aims to address this issue by providing a framework that enables pattern matching between a software log and an event pattern expression that is based on a monitoring policy. The framework consists of parsers and annotators that facilitates transformation of a monitoring policy into a Petri Net as well as source code annotation for gathering data through logged events. It further expands upon this work by proposing an adaptive logging framework that will greatly improve the quality of log management by autonomically adjusting the amount of information logged based on the application’s operational environment. Finally, a prototype system of the policy driven monitoring framework is implemented and tested with applications of different scales as a proof of concept for the proposed framework.