A Systematic Review of Logging Practice in Software Engineering
Abstract
Background: Logging practice is a critical activity in software development, which aims to offer significant information to understand the runtime behavior of software systems and support better software maintenance. There have been many relevant studies dedicated to logging practice in software engineering recently, yet it lacks a systematic understanding to the adoption state of logging practice in industry and research progress in academia. Objective: This study aims to synthesize relevant studies on the logging practice and portray a big picture of logging practice in software engineering so as to understand current adoption status and identify research opportunities. Method: We carried out a systematic review on the relevant studies on logging practice in software engineering. Results: Our study identified 41 primary studies relevant to logging practice. Typical findings are: (1) Logging practice attracts broad interests among researchers in many concrete research areas. (2) Logging practice occurred in many development types, among which the development of fault tolerance systems is the most adopted type. (3) Many challenges exist in current logging practice in software engineering, e.g., tradeoff between logging overhead and analysis cost, where and what to log, balance between enough logging and system performance, etc. Conclusion: Results show that logging practice plays a vital role in various applications for diverse purposes. However, there are many challenges and problems to be solved. Therefore, various novel techniques are necessary to guide developers conducting logging practice and improve the performance and efficiency of logging practice.
Type
Publication
In Asia-Pacific Software Engineering Conference
Authors
Authors
Authors
Authors
Shenghui (Samuel) Gu
(he/him)
Postdoctoral Researcher
Postdoctoral Researcher at the University of Ottawa specializing in Trustworthy AI and Software Engineering, holding a Ph.D. from Nanjing University.
Research lies at the intersection of AI Safety and System Reliability, with deep expertise spanning LLM-driven testing, search-based software engineering for autonomous systems, and AIOps for distributed architectures.
Dedicated to developing rigorous, interpretable, and scalable methodologies that leverage generative AI to solve complex validation challenges in safety-critical and large-scale industrial systems.