Cloud-Native Engineering Simulation Framework for High-Performance Computing
Abstract
Cloud-Native Engineering Simulation Framework for HPC presents a new approach to compute complex engineering simulations in cloud-native technology and high-performance computing (HPC) resources. Standard engineering simulation systems frequently struggle with scalability, costly infrastructure requirements like memory bandwidth and other resources and automation of computational workflows. This framework proposes a solution for these challenges by integrating containerization, microservices using orchestration, distributed computing and elastic cloud resources to have an all-in-one simulation environment. This framework allows engineering applications to allocate computational resources dynamically and as per workload needs, thus achieving large improvements in simulation speed, scalability/flexibility and utilization of computing resources. Cloud-native principles also enable automated deployment, fault tolerance, as well as workload management and straightforward integration of simulation tools and data-processing pipelines. The methods that we proposed can be used in computationally intensive domains such as: (i) structural analysis, (ii) fluid dynamics and heat transfer simulations [26], impact prediction of aerospace structures over operational conditions from processing history [19], including recent efforts in digital twin applications. In short, the framework allows running of engineering applications in a high-performance setting while also minimizing infrastructure dependence and supporting resource usage enhancements performed over virtual machines. The framework also facilitates use of advanced technologies like artificial intelligence, machine learning and digital twins along with high-performance engineering simulations in a single Integrated-Cookbook approach using highly available real-time data. Utilizing cloud-based data storage and distributed computing environments enables the efficient processing of large-scale simulation data in a live or near-real-time mode. Containerized simulation environments offer platform independence and automate deployment, setup and maintenance of engineering applications. Also, with automated workflow management, less human involvement is needed and it also provides scientists and engineers an opportunity to run multiple simulation experiments at one time. The new cloud-native framework enables greater sharing so that sim resources and comp workflows are accessible to delivery teams in remote locations. Its elastic structure helps in scaling up and down of resources as per computational workload, thus optimizing performance and saving unnecessary infrastructure costs. Therefore, it can facilitate both small experiments and trials as well as large industrial simulations. Integrating HPC features with cloud native architecture will provide a scalable and efficient computational environment for the next generation of engineering simulation that can facilitate faster decisions, better resource utilization and more correct outcomes in Engineering.