Accelerating Product Engineering Using AWS Solutions with Vitesco Technologies

Headquartered in Regensburg, Germany, Vitesco aims to power clean mobility by designing and manufacturing automotive parts for electric and hybrid vehicles. To test out new ideas and products, its engineering teams run complex, large-scale simulations and manage the development through PLM software.

Vitesco originally ran its simulations with on-premises resources and clusters. To support a broad product portfolio and its electrified vehicles strategy, Vitesco wanted to run more simulations at faster speeds. The company faced the decision of either procuring more hardware or working around bottlenecks in its engineering workflows.

“In the past, we had clusters with a fixed number of cores for our simulations,” says Theodor Heindl, head of Engineering Applications at Vitesco. “We had a first come, first served principle, which made it difficult to prioritize urgent simulations. The other option was to build a big simulation farm that would be running day and night, whether we needed it or not.”

On spinning off from Continental AG in 2018, Vitesco needed to build infrastructure for over 350 mission-critical applications, including its PLM software, in 18 months. “We were working against a tight time frame,” says Alwin Perras, service owner for 3D mechanical simulation tools and HPC at Vitesco. “We needed a dynamic system that would help us review and control compute costs.” These requirements led the Vitesco team to architect an HPC environment for running its engineering and simulation workloads on AWS. Vitesco developed a proof of concept in 3 days by working alongside AWS Professional Services—a team that helps businesses achieve their desired business outcomes using AWS.

To quickly get its systems running on AWS, Vitesco migrated its data records using Amazon FSx for NetApp ONTAP, a fully managed shared storage built on NetApp’s popular ONTAP file system. Next, Vitesco’s HPC team worked on modernizing Vitesco’s on-premises clusters with cloud-based and HPC solutions, including AWS ParallelCluster, an open-source cluster management tool that makes it simple for businesses to deploy and manage HPC clusters on AWS. Using cloud-native services, Vitesco can respond better to business needs by automatically scaling its infrastructure up to 4,100 cores. This is five times the number of cores that Vitesco had in its on-premises environment.

To provide centralized data storage to its engineering teams, Vitesco stores its data in Amazon Simple Storage Service (Amazon S3), an object storage service built to retrieve virtually any amount of data from anywhere. As a result, engineers save precious time because they don’t have to download simulation results to their computers. They can also securely share and collaborate on their results from nearly any location. Across the company, Vitesco has stored more than 3 PB of data on AWS. “Our system supports agile development and engineering,” says Perras. “Our teams can exchange data and results in one central environment without having to think about it.”

In a matter of minutes, Vitesco’s engineers can now spin up the resources that they need, such as Amazon Elastic Compute Cloud (Amazon EC2) G4 Instances and Hpc6a Instances for secure and resizable compute capacity for virtually any workload. Vitesco’s engineers can choose their instance types based on the urgency of the project and estimated cost. For example, they can run simulations with compute instances, such as the Amazon EC2 Hpc6a Instances, and perform 3D rendering with graphics instances, such as the Amazon EC2 G4dn Instances. In turn, Vitesco has saved 55 percent on compute costs compared with operating on premises. “Our global teams really need speed of development and simulations to deliver business value,” says Heindl. “AWS provides huge flexibility and the ability to ramp up our infrastructure and try out new ideas.”

Unlike in the on-premises environment, Vitesco’s engineers can run simulation jobs simultaneously without causing bottlenecks in their workflows. The runtime of their simulations has also improved. Simulations that used to take 1 week are now delivered in 1 day, helping the company ship new products to customers faster than before. “One of the biggest benefits we’ve experienced is speed,” says Heindl. “Our teams can immediately derive insights from their simulations to further develop our products.”