DevOps Driven Software Performance Assurance for Large-scale Software Systems

点击翻译按钮获取中文摘要

The rise of large-scale software systems (e.g., Amazon Web Services) has become an important role in people's daily lives. Failures in such systems are more often associated with performance issues, rather than with feature bugs. Therefore, performance assurance activities are an essential step in the release cycle of large-scale software systems. Performance assurance activities aim to identify and eliminate performance issues during the development and operation of such large-scale software systems. Examples of performance issues are response time degradation, higher than expected resource utilization and memory leaks. Such issues may compromise the user experience, increase the operating cost of the system, and cause field failures. Failure in detecting such issues would result in significant financial and reputational repercussions. Performance assurance for large-scale software systems is facing great challenges in the era of DevOps. On one hand, detecting, locating, understanding and fixing performance issues are often conducted at a late stage in the development circle, with large amounts of resources required. On the other hand, the fast-paced release cycles of DevOps may not permit enough resources (e.g., long running time and dedicated environment) for the costly performance assurance activities. Moreover, existing approaches for performance assurance activities do not bridge the knowledge gap between software development and operations, which is of great significance in DevOps. The goal of the proposed research is to provide practitioners with novel software performance assurance approaches for large-scale software systems in the era of DevOps. To achieve our goal, I plan to improve the practice of performance assurance in order to be easily adopted by the engineering of large-scale software systems that follow DevOps practices. In particular, I propose to proactively ensure software performance during software development, such that practitioners can improve software performance and address performance issues in a timely and cost-effective manner. In addition, I propose to address software performance issues directly during the operation of the software in the field, with the end users. Finally, I propose the design of a systematic infrastructure and an analytical framework that bridge the knowledge of software performance with information uncovered from both the development and operation. Large-scale empirical studies will be performed on large open source and industrial systems, to understand the benefits and limitations of our work. The proposed research is a pioneering attempt to advance the practice to ensure the performance of large software systems that serve millions of users worldwide. Furthermore, the proposed research will expose, train and enable highly qualified personnel (HQP) to contribute to the state-of-the-art in software engineering research.