BWT Alpine F1 Team turns big data into faster track times with data science platform built on Azure and Azure NetApp Files

An entire organization with a single goal

After working with Microsoft to move IT operations from the datacenter to Microsoft Azure, Alpine IT Business Systems and Data Science Director Nathan Sykes, says, “We access data in Microsoft Azure at a speed faster than we did on site. The performance of the cloud is not a limiting factor.” Though, he adds with a smile, “The speed of light is now becoming a bit of a factor, and we're having problems fixing that one.” 

BWT Alpine F1 Team drivers may not approach the speed of light; they have a more modest goal of remaining quicker than their opponents. In that, world-class driving skills are augmented by the efforts of their IT colleagues since, as Vice President of IS/IT Pierre d’Imbleval confirms, “The entire organization is here solely to make the car go faster.”

Winning with a data science platform—and very fast cars

The team already recognizes the benefits of migrating business operations to Azure, and had previously invested in Microsoft Dynamics 365. Now, for the 2022 season, the team has built its new cloud-hosted Alpine Data Science Platform that uses additional Azure infrastructure to bring racing, testing, and manufacturing data together into a powerful, comprehensive solution that helps turbo-charge winning performance. 

Sergio Rodriguez manages Data Science and Data Engineering, and he points out that, for him, among the most important drivers in Formula 1 is data. “BWT Alpine F1 Team is a data-driven company. We cannot make decisions without data.” That includes data from the cars on the track, he says: “Our car is a very big IoT device with hundreds of sensors, thousands of channels. We generate around 3 million samples per second in one car.” And by the way, he adds, “We have two cars.” The testing facility in France generates data as well, as does the chassis design and production operation in the United Kingdom. And there’s all that data generated and consumed by manufacturing and by the entire end-to-end logistics infrastructure that gets new and reworked parts designed, built, and installed on the cars within 24 hours. 

“We had to have simple files located on-premises, so we would copy and paste them across the network to wherever they might be requested,” says Rodriguez. He identifies four key requirements for the new platform. “One is data capture. We need to capture billions of data points sending in bursts of data. Second element is data store. We need to store the data securely, with rapid retrieval. Third is presentation, customized for users through different APIs, Power BI dashboards, apps, and services. And the last big part of our data science platform is how our engineers design and run AI models on top of that data to provide new insights.” In meeting these needs with the Azure-based data science platform, BWT Alpine F1 Team has revolutionized workflows and helped put itself in pole position for the 2022 season.

From Platform to Podium, using Azure infrastructure

Rodriguez says his choice of foundation for the new platform was easy. “The BWT Alpine F1 Team Data Science Platform is built entirely on the speed and security of Azure NetApp Files and the Azure cloud platform.“

Huge volumes of data are piped into groups of clustered virtual machines (VMs) by Azure Batch, from high-performance, highly scalable Azure NetApp Files file shares. “We need to present the data fast to our engineers at track, or in the factory. Azure NetApp Files provides encryption, highly secure connections—but above all, it gives us that speed; it’s the perfect technology for us, “ says Rodriguez. BWT Alpine F1 Team found an ideal combination in Azure VMs and Azure NetApp Files that offers performance, reliability, and high availability. Each cluster handles specialized data processing for data sources such as testing, car performance monitoring, simulators, wind tunnels, and manufacturing in a KX data refinery running natively on Azure. Proximity placement groups help speed data flow, while clusters can be dynamically re-sized with additional VMs on demand, without interruption to data processing as new nodes are served by the same independent Azure NetApp Files endpoints. Rodriguez says, “By necessity we keep a fixed number of resources on-premises, but the connection with our VMs in the cloud is transparent. So, we are free to grow horizontally with more virtual machines for our engineers, for example, and also we can scale vertically, adding GPUs for our AI training, a terabyte of RAM; it’s pretty straightforward to scale up whatever we need.”

There are currently more than 265 virtual cores, 7 terabytes of RAM, and 65 terabytes of storage in a platform that’s hungry for data. It’s fed by high-speed data lanes from those on-premises and in-car sources into the cloud, via Azure ExpressRoute, which takes advantage of the speed and resiliency of private Microsoft dark fiber connections. The result, Sykes says, is that “Data ingestion from track to the data platform and back to the track for analysis is faster than a TV signal, with times under 1 second. It’s near real-time.” Here, as with the cars, speed is crucial to BWT Alpine F1 Team’s operation. “Don’t talk so much about big data in Formula 1. Talk about fast data. Fast is what matters for us,” says d’Imbleval. 

The value of that data in its contribution to performance lies in the speed with which it can be gathered, stored, published, and accessed. “Azure NetApp files allows us to retrieve the data fast and in a secure way and that’s very important for us because we need to make decisions in fractions of second,” Rodriguez adds. Stale, unavailable, or invisible data is just that: data, instead of information that can make invaluable contributions to the overall operation. 

Speed, power, and efficiency

Cost is, of course, also a consideration even in the glamorous world of Formula 1. Or, rather, return on investment—particularly in the last couple of years, given the Formula 1 organization’s introduction of cost caps as a way to prevent unlimited spending. In 2022, the cap reached $140 million, and for d’Imbleval, that just underscored the need to get the best mileage from his IT investments. “We need to reinvent the way we work to increase our productivity with the same level of resources,” he says. Alpine can achieve this with Azure, says Data Engineering Manager Rodriguez. He describes how, with the Azure infrastructure, Alpine not only achieves availability and scalability, it delivers reliable results, reduces capital expenditure, and squeezes the most performance out of those capped costs. “We need a high level of flexibility to rapidly scale resources up and down as needed, and in Azure that’s really straightforward. During a weekend, we spin up a lot of compute power and associated Azure NetApp Files resources for engineers to collect data, run their models, and analyze performance. Once it’s all over we can scale down everything until the next race,” says Rodriguez. 

d’Imbleval sums up his experience building the BWT Alpine F1 Team Data Science Platform with Azure. “We develop a car for performance first. But once we achieve maximum performance, then we need ensure reliability. So, there’s a perfect similarity in how we develop our car and in how we developed our IT infrastructure with NetApp Files on Microsoft Azure.”

Find out more about BWT Alpine F1 Team on Twitter, Facebook, LinkedIn, and YouTube.