A modular simulation-based MBSE approach applied to a cloud-based system

cloud-based system estimation;model-based systems engineering;time-base simulation;modular modeling approach

Data systems consist of a network of communication channels, applications that trans-
mit data across these channels, and the hardware running these applications or generating the data. Most modern data systems include cloud storage or compute which has unpredictable or stochastic properties making estimations of cloud behavior and performance difficult. Resource usage is function of behavior and performance on software/hardware. Cloud cost is a function of resource usage and hardware used. Public cloud spend was over budget by an average 18% for 2022 with organizations reporting an estimated 28% public cloud waste. The scale of this problem is a measure of the difficulty of accurate cloud-based system performance and cost predictions. The goal of this paper is to develop and demonstrate a modular and scalable Model-Based Systems Engineering (MBSE) approach for designing, updating, and managing cloud-based data systems. Our use-case based Agile MBSE approach is developed to integrate with commonly used Agile software development processes to increase collaboration between system engineers and developers. We embed simulation behaviors within the lowest level of system specification activities to produce a modular and
reusable set of simulation-ready system activities. Our approach uses a combination of languages (SysML, fUML, Apache Groovy, and the Action Language Helper (ALH)) to develop these modular system activities for scalability and speed. We applied this approach to the simulation of a cloud-based data system. The results show that our approach produces a modular, time-dependent, executable system model that can estimate cloud-based system performance and storage cost as a function of time. Emergent behavior observed from the simulation results indicate that the system
model is capable of providing system engineers and management teams valuable insight into the behavior of the system they are designing or upgrading.