Li Auto runs on Amazon Web Services to reshape the user's Internet of Vehicles experience

To meet the business needs of its domestic market, Beijing-based Li Auto, which designs, develops, manufactures, and sells premium smart electric vehicles, has been focused on building a flexible, resilient, and reliable Internet of Vehicles platform. Li Auto’s Internet of Vehicles system consists of three parts: a vehicle terminal (vehicle transmission control unit), a cloud computing processing platform (vehicle cloud), and a data analysis platform (big data). It serves users in Xinjiang, Tibet, Qinghai, Gansu, Ningxia, and Shaanxi provinces.

As its core business system, Li Auto’s Internet of Vehicles platform provides functions and services throughout the vehicle’s life cycle, using the cloud to process and analyze data, such as vehicle and electric energy information. The platform also filters and cleans the data for further mining, to aid car owners and Li Auto’s business team.

When it first started, Li Auto chose to build a cloud native operations service platform as it saw a future of intelligent networked vehicles. However, as its business and users grew, the Internet of Vehicles platform faced challenges. These included network jitter, cloud disk failure, and insufficient cloud resource application. Its existing computing, storage, and network performance could not scale in some cases. Li Auto had to secure a large number of idle resources in advance, while sometimes it experienced network latency, and could not respond quickly to an owner's request. Li Auto’s own intelligent cloud control platform was connected to an existing local cloud provider through an API, but some functions were not supported, and it had to intervene manually.

"The efficiency, safe access, and the vehicle-to-cloud round trip time (RTT) network latency directly affect the user's driving experience," said Zhang Yongxin, director of Li Auto Intelligent Cloud Platform. "We hope to work with a new cloud service provider to improve the platform’s performance and latency for our users. This includes improving the robustness and stability of the system, networking flexibility, and meeting security and compliance requirements."

Li Auto conducted a comprehensive evaluation to select a new cloud service provider, including looking at technology leadership, product maturity and reliability, cost performance, and service responsiveness. With Li Auto’s business growing rapidly, every second its business is disrupted was unacceptable. It needed a stable infrastructure to allow them to focus on user service instead of operations and maintenance, and realize savings on human resources and other non-resource costs. Li Auto chose Amazon Web Services.

"We wanted to bring in a market-leading supplier who was closer to our global end-users to provide them with a high-performance, low-latency connected car interaction experience, while significantly improving the robustness and stability of the system, the flexibility of networking, and enhancing security compliance requirements," said Zhang.

By running on Amazon Nitro, Amazon Elastic ComputeCloud (Amazon EC2) and Amazon VPC, Li Auto stabilized its basic platform services, eliminating network jitter and unexpected service interruption. It also used Amazon Graviton2 processors and Amazon Elastic KubernetesService (Amazon EKS) to deliver the best price performance for its cloud workloads running on Amazon EC2, and lower costs without modifying code or worrying about compatibility. Its services relied on Amazon Aurora and Amazon OpenSearch Service for database management, logging and other needs.

"The automobile is a data producer. We use Amazon Simple Storage Service (Amazon S3) for data collection, processing, handling, or analyzing complex, big data and AI applications for the vehicle, in order to achieve an efficient data exchange,” said Zhang. In addition, Li Auto has improved cloud API management with Amazon SAM CLI, SDK, CDK, Amazon CloudFormation and Terraform.

To provide safe and stable Internet of Vehicles services for car owners, Li Auto had to update its security architecture. Li Auto used Gateway Load Balancer to easily deploy, scale, and manage third-party virtual appliances. That provides the company a gateway for distributing traffic across multiple virtual appliances while scaling them up or down, based on demand. The gateway decreases potential points of failure in its network and increases availability. At the same time, Li Auto used a third-party network security firewall to achieve east-west and north-south network traffic control. The Company also used Amazon CloudFormation to manage the multi-VPC deployment of Amazon Transit Gateway to connect VPCs and on-premises networks through a central hub. This connection simplifies its network and puts an end to complex peering relationships.

Amazon Web Services trained Li Auto’s team during deployment. "The training was comprehensive. It not only covered cloud concepts, but also included on-site practices. This combination of knowledge and practices, and the immersive experience, allowed us to gain an in-depth understanding of various cloud services we use frequently,” said Zhang.

Li Auto further worked with Amazon Web Services' Cloud Economic team to analyze and optimize how to better use Amazon Compute Optimizer to optimize resources and Amazon Elastic Block Store (EBS) to improve cost performance.

After running on the Amazon Web Services, Li Auto has seen significant improvements in reliability, performance, and the security of its Internet of Vehicles platform. The network RTT delay has been reduced from more than 100 milliseconds to fewer than 60 milliseconds Performance has improved by 40 percent. Basic network performance has improved 30 percent.

Looking forward, Li Auto aims to "introduce mature cloud services that better fit the application scenario and use cases efficiently". The company is evaluating expanding the use of Amazon EKS across more scenarios, including with Spot instances and Karcenter elastic scaling. It is also looking at standardization and cross-platform capability improvement of cloud native, serverless and Function-as-a-Services (FaaS)-related services. This will link the cloud edge and provide better service and experience for its car owners.

Li Auto was founded in 2015 and headquartered in Beijing. The company is committed to creating a safer, more convenient and comfortable smart electric car for families, and become the preferred luxury electric car brand for them. Its first vehicle, Li Auto ONE, launched in October 2018, has sold more than 200,000 vehicles since December 2019. In 2022, Li Auto debuted a six-seater SUV, Li L9, and a six-seater luxury SUV, Li L8. With the launch of its new flagship, Li L7, in February 2023, Li Auto has fully covered the diverse needs of families for luxury electric vehicles. Li Auto operates 296 retail centers nationwide, covering 123 cities to date.