Click to learn more about author Daniel Foulkes Leon.

Speed and power are driving the adoption of NoSQL technologies in Motorsports. When it comes to racing and speed, the objectives of Formula 1 and NASCAR teams are the same. Winning. And they aim to do so by having the fastest car, recruiting the most skilled drivers, and making the best strategic and competitive decisions both before and during the race. Enter the era of gargantuan data volumes that are growing exponentially: from a plethora of sensors within the cars, the drivers’ uniforms, trackside scanners, video feeds, weather reports, you name it. Database technologies and their management are now factors that can impact not only the decisions happening at the track, but they are also the key to securing that podium finish.

Formula 1 and NASCAR derive clear benefits from data analysis and processing, but their implementations and approaches are very different from one another. Within Formula 1, the team’s telemetry data is proprietary and highly guarded. NASCAR, on the other hand, makes telemetry data from all vehicles and teams available in real-time to fans, teams, and OEMs (Original Equipment Manufacturers, which are Ford, Chevy, and Toyota for NASCAR). But both Formula 1 and NASCAR share one key element: the rapid adoption of NoSQL technologies.

Augusto Cardoso is the Lead Engineer at SportMedia Technologies (SMT), the partner with NASCAR that aggregates, processes, and transmits all race data to various audiences. Cardoso indicated the pace of their adoption is growing across multiple sports:

“We first adopted MongoDB in the Motorsports group, starting around three years ago. It replaced a SQL database. We are expanding our use of MongoDB in other sports like hockey and baseball. Each sport has its own requirement, and MongoDB allows for great flexibility.”

Phillip Thomason, Lead Engineer at the British Formula 1 racing team and constructor for Williams Racing, identifies two big advantages of using NoSQL technologies: breaking the walls between previously siloed repositories and allowing for collaboration across multiple teams. NoSQL helps solve particular problems that would typically occur when silos are well dug in. Thomason describes what it was like before they began using NoSQL technologies, “These queries were slow, manual, and often practically impossible. NoSQL has allowed users to access the potential of all data and improved inter-departmental communication.”

In this data-rich environment, embracing the variety and diversity of data through a “right tool for the job” philosophy has allowed NoSQL technologies to provide the power and speed that these teams require.

This is no easy feat since Formula 1 teams generate at least 3 TB of data per race, and over 100 million data points alone are created in a single weekend by NASCAR.

Data is transmitted from the cars to the base stations across the track, and tools need to process this fast, both for analysis by the teams as well as for video animation pairing for broadcasting.

“In the case of NASCAR, the data needs to be ready for multiple audiences in real-time,” Cardoso says, continuing:

“In Motorsports, we have some unique applications where we have a production truck that travels to each venue, weekly. We provide services for at-track users, including teams, car manufacturers (OEMs), and TV broadcasters. The challenge is that we need to have data replicated in both the truck and in the cloud. Customers use the data both directly from our truck or from the cloud. In case of an internet issue, the TV broadcast can’t stop. Our local infrastructure can operate independently, but the majority of the performance analysis and Data Science users connect to the cloud.”

High-octane data performance is an essential part of the business, but it can’t come at the cost of usability. Cardoso adds, “since we moved to MongoDB, with an ‘in the truck’ and cloud presences, we haven’t had a single database related issue.”

Within F1, teams are actively trying to gain a competitive advantage over one another, and their use and combination of various technologies is no exception.
         
“F1 teams are obliged to develop much of the required software ‘in-house’ as it is simply not available on the market,” explains Thomason, continuing:

“As we’re a relatively small team, technology choice is often driven by existing skills within the team. Often, there is insufficient time (or resources) to recruit additional skills. We rely on very capable team members that all work on the ‘full-stack’ and day-to-day their roles involve research into new tech.”

Within Williams, the process of bringing together structured and unstructured data is a key competitive area and one where not much else can be publicly disclosed.

“NoSQL has allowed users to access the potential of all data, and it’s also improved inter-departmental communication.” Phillip Thomason, Williams Racing

As database technologies advance in potential and opportunities, so do methods. Motorsports is no exception in how the push for DevOps processes are leading the way for Continuous Integration-Continuous Development (CI-CD). On DevOps processes, Thomason added that for Williams in F1:

“We are always faced with time and resource challenges and a drive to improve efficiency across all areas of the business. Tools like Docker have allowed us to move the traditional software development team towards the DevOps arena and helped to better define the boundary between IT and Software departments. We always had a very fast software update mechanism (releasing on a race-to-race basis), but DevOps has certainly given us more flexibility.”

Telemetry data is not the only type of data that can be gathered during a race; other sources can also have huge impacts on the teams and their bonuses. One such example is within Pit Stops. Since the banning of mid-race refueling in Formula 1, their Pit Stops are now consistently in the sub-three-second mark, with Williams achieving the fastest Pit Stop in 2016. The push for shorter Pit Stops is also seen within NASCAR, where teams can track and monitor the Pit Stop times of all cars during the race. This has led teams to award special bonuses to their teams based exclusively on the data aggregated by SMT.

NASCAR and SMT process their data entirely through MongoDB and microservices. Cardoso even indicated that they hadn’t had a single production failure since they implemented them:

“The microservices are so much quicker. I already have a funnel with scripts and everything set up … So in terms of CI-CD, it’s much easier for me to keep rolling these things out … and I can just add it in and edit.”

On the adoption of data, Thomason added:

“F1 has always been data-driven, and with the governing body restricting what we can do with testing (limited tire, computational fluid dynamics, wind tunnel, and track testing), this had led to an efficiency drive around the technology to extract the maximum benefit from the running we are allowed. Telemetry data expansion was therefore driven by the teams seeking a competitive advantage.” 

Whereas the two sports place a high value on data, and both reap clear rewards from its analysis, their approach is very different. For Formula 1, Thomason describes it as a highly valuable and coveted resource:

“Given the history of F1, any data publicly released would be jumped on by teams to analyze their competitor’s performance and would require significant investment in resources for teams to remain competitive.”

NASCAR, on the other hand, is strikingly different. “That’s one thing NASCAR did,” Cardoso added, “everybody can see everybody’s data. So that is a very important thing. And that was all NASCAR policy! For that, I commend them because that makes it more accessible.”

This has led to a very different approach to the data and how competitiveness is built around it. Cardoso is part of a team that developed an application used directly by NASCAR, the teams, and its OEMs that helps calculate the fuel efficiency and estimate of every car in real-time. The application runs instantaneously. When showing the speed of the results, Cardoso added, “I don’t know if you can appreciate how fast this was. This aggregation there is 28 stages.” Not only can a trailing team see in real-time if the cars in the lead have enough fuel to finish the race, but fans and competitors too can strategize along as the race develops.

Formula 1 and NASCAR might treat telemetry data differently, but both want their data to be transferred as fast as possible. Data transfer rates are the general bottleneck for both F1 and NASCAR, as issues with connectivity can happen due to network errors or base stations that flood alongside the track during rainfall. Even so, they have achieved impressive latency speeds already. Within NASCAR, SMT is able to process the data from the track to their on-prem truck, to the cloud, and to their users even faster than the general broadcast delay. Cardoso said:

“So just to give you an idea, when you are at the track, the latency to the data that we send is about 5 to 6 milliseconds. It’s really fast. The data latency when it gets to the cloud, it gets to be closer to 100 milliseconds. And that big gap, basically going from the truck to the data center, the fastest I can get that is about 60 milliseconds of latency.”

The push for faster data speeds is also shared by Williams, as Thomason indicated:

“The drive is to move processing as close to the race as possible in order to reduce latency in the data processing/stream enrichment suite. A tenth of a second saved here has a real race-performance benefit.”

Speed, power, and versatility are the requirements of every element of motorsports, from aerodynamics to tire quality and telemetry streams. Database technologies and architecture models are no exception. NoSQL is now another reliable element to help gain that next fraction of a second.