The public defence of the thesis takes place on Friday November 8th, 2024 at 09:15.

Thesis title: Enhancing DevOps with Autonomous Monitors: A Proactive Approach to Failure Detection

Author: Adha Hrusto, Department of Computer Science, Lund University

Faculty opponent: Professor Mika Mäntylä, University of Helsinki, Finland

Examination Committee:

• Professor Helena Holmström Olsson, Malmö University
• Doctor Sahar Tahvili, Ericsson and Mälardalen University
• Associate Professor Deepika Badampudi, Blekinge Institute of Technology
• Deputy: Associate Professor Mikael Nilsson, Lund University

Session chair: Associate Professor Elizabeth Bjarnason, Lund University

Supervisors:

• Professor Per Runeson, Lund University
• Associate Professor Emelie Engström, Lund University

Location: E:A, E-building, Klas Anshelms väg 10 / John Ericssons väg 2, Lund, Sweden

For download: Follow this link to download the pdf: Link to follow...

Abstract

Software engineering practices, including continuous integration, continuous testing, and continuous deployment, aim to streamline and automate the software development process. A cultural and professional movement that builds upon continuous practices, DevOps, seeks to bridge the gap between development and operations. By fostering a collaborative environment, DevOps supports faster, more frequent, and reliable software releases, inherently promoting agile methodologies throughout the software development lifecycle.

By introducing agility, there is a higher risk of operational failures in cloud-based software systems. Recognizing this challenge, the objective of this thesis is to understand and present approaches for mitigating the cascading effects of operational failures across interconnected system components. In collaboration with two Swedish companies, we investigated how proactive monitoring strategies inspired by state-of-the-art machine learning (ML) solutions can prevent failure propagation and ensure seamless system operations.

The conducted research activities span from practice to theory and from problem to solution domain, including problem conceptualization, solution design, instantiation, and empirical validation. This complies with the main principles of the design science paradigm mainly used to frame problem-driven studies aiming to improve specific areas of practice.  

The main contributions of this thesis are threefold. First, an in-depth overview of operational challenges and matching solutions in cloud-based software systems, focusing on alert management and monitoring data through two case studies and extensive literature reviews. Second, a proactive alert strategy called autonomous monitors to enhance early detection and prevention of operational failures. Finally, the practical applicability of these monitors is confirmed via empirical studies, highlighting their effectiveness in various industrial contexts.

We demonstrated the practical effectiveness of the proposed ML-based monitoring solution to pave the way for its widespread adoption for enhancing DevOps.