Predicting Failures in Dynamic Software Systems
Staff - Faculty of Informatics
Date: 15 December 2023 / 15:30 - 17:00
USI East Campus, Room D0.03
You are cordially invited to attend the PhD Dissertation Defence of Rahim Heydarov on Friday 15 December 2023 at 15:30 in room D0.03 (USI East Campus).
Runtime failures of evolving complex software systems are unavoidable due to their intrinsic characteristics. Moreover, traditional development-time testing methodologies are extremely limited in coping with runtime-emerged issues. However, prediction and localization of failures in the ?eld help mitigate their negative impact by enabling self-healing mechanisms currently available in modern industrial solutions. Nevertheless, the state-of-the-art approaches mainly rely on models that do not cope with emerging faults or are ineffective in dynamic execution conditions. In this PhD, we aim to overcome the main limitations of the current methods for predicting and localizing failures by defining approaches that do not require a long supervised training stage, work in dynamically evolving execution contexts with highly variable metric sets, and cope with various common failure types. We present the first complete approach that predicts failures and localizes failing components in large-scale auto-scaling distributed applications. We present an extensive data set collected by running distributed cloud applications widely used in commercial environments, which we provide in a replication package for further studies. We present the results of a comprehensive comparative evaluation of the effectiveness of the defined approach on the collected data by experimenting with a set of common failure types.
- Prof. Mauro Pezzè, Università della Svizzera italiana, Switzerland (Research Advisor)
- Prof. Giovanni Denaro, University of Milano-Bicocca, Italy (Research co-Advisor)
- Prof. Luca Maria Gambardella, Università della Svizzera italiana, Switzerland (Internal Member)
- Prof. Paolo Tonella, Università della Svizzera italiana, Switzerland (Internal Member)
- Prof. Davide Taibi, Tampere University, Finland (External Member)
- Prof. Damian Tamburri, Eindhoven University of Technology, The Netherlands (External Member)