In Search of Optimal Solutions: ADRO’s Migration to AWS

Founded in 2020, ADRO is a startup specializing in the production of high-quality body kits with an aerodynamic design. The company possesses advanced technology that integrates aerodynamics into car tuning body kits, aiming to optimize driving performance. ADRO CEO Yoon Seunghyun highlights the positive reception in overseas markets, attributing it to the incorporation of technologies such as lightweight carbon fiber and drag reduction into car body kits, which traditionally focused solely on design. This success has propelled ADRO's expansion into global markets, including North America, Australia, the Middle East, Europe, and Asia, resulting in a notable increase in sales.

To enhance the quality of its body kits, ADRO applies Computational Fluid Dynamics (CFD) technology. However, limitations with their high-end PC hindered efficient CFD simulations, with simple simulations taking days to complete. The existing infrastructure lacked the scalability and agility required. Furthermore, a partnership with the Kumho Tire Ecsta Racing Team resulted in a surge in analysis volume, necessitating a solution for faster simulations to analyze airflow during car driving. ADRO sought a scalable and flexible infrastructure solution that also offered cost efficiency.

Simulations aimed at optimizing aerodynamic designs demand high-performance computing environment, a limitation faced by ADRO due to constraints with their existing a high-performance computing environment. Acquiring the necessary equipment was both costly and time-consuming, prompting the consideration of cloud solutions that allow the utilization of the highest-spec infrastructure when only required. “After evaluating multiple cloud vendors, AWS stood out as the only platform that met our requirements for running CFD simulations. Additionally, the AWS Activate Startup Program not only provided us with essential credits but also offered valuable technical support. Implementing CFD for body kit development would have been considerably challenging without the robust capabilities and support from AWS,” said ADRO CEO Yoon Seunghyun.

To cluster multiple instances for running CFD workloads, ADRO uses AWS ParallelCluster. For parallel processing, ADRO clusters multiple Amazon EC2 Hpc7g instances based on AWS Graviton processors and utilize OpenFOAM, an open-source CFD solution, within their AWS HPC cluster environment. Data storage is configured using Amazon FSx for Lustre, a fully managed service providing the parallel filesystem Lustre. “This strategic use of AWS's HPC solutions resulted in a remarkable reduction in analytical simulation time, from three days to three hours,” said the CEO.