Devops for digital business: Optimizing the performance and economic efficiency of software products for digital business

Abstract

Software is playing an increasingly crucial role for any modern enterprise. Due to the high volatility of software and its environment, management of software change has become a crucial ability of a digital enterprise. Indeed, the ever-changing technical aspects of cloud-based software might influence its economic aspects at the same rate. This PhD work proposes an autonomic management framework for economic-driven adaptive resource allocation, during which performance and economics are considered simultaneously. The framework consists of a hybrid performance-economic model for cloud software systems and a self-adaptive system for resource management at run time. The proposed framework will be developed and evaluated on use cases from the domain of financial services. The objective in these use cases is to improve productivity of data scientists when they train and deploy analytical and ML models, while maintaining a low resource consumption. Considering productivity and cost as the economic factors, and latency and resource consumption as the technical ones, we will use the proposed hybrid model to guide initial deployment, and the self-adaptive system to adjust the deployment at run time.