Red Bull Racing’s Malaysian win and its Milton Keynes IT

Red Bull Racing came first and second at the Malaysian Grand Prix. By chance I was among a handful of analysts and journalists invited by one of its technical partners IBM to visit its HQ in Milton Keynes on the Friday race day. You’ll be interested to learn more about the team’s challenges and how it uses IBM and other suppliers’ IT to overcome them. Formula 1 can help other industries to address efficiency and performance, while it illustrates some of the key features of IBM’s Spectrum offerings.

Red Bull Racing’s challenges

Matt Cadieaux (CIO) and Al Peasland (Head of Technical Partnerships) gave an overview of the challenges to the team and the applications they run. The team’s wind-tunnel (used to test 60%-sized cars) is based in Bedford – otherwise its resources are based in Milton Keynes, which includes 8 buildings to house the 700 employees. IT is connected temporarily via MPLS to each of the (in 2016) 21 racetracks around the world in turn. Alongside IBM, AT&T is also a technology partner and we got to look inside the new control room, which has a very active role during practice sessions, qualifying and the race itself. Al summarised the challenges as:

• The global nature of the sport – currently there are 21 races in a year with a new track added each year on average. In 2016 the track in Malaysia had been resurfaced and some of its corners resurfaced, making preparations different from before. In total there are around 500m viewers of Formula 1 races.
• Compliance with technical and sporting regulations – in 2016 these were similar to 2015, although Pirelli added a new tyre compound, 2 races were added and testing time was reduced to only 8 days before the first race. Testing aerodynamics is a Formula 1 team’s most heavily regulated workload, with wind tunnel usage and CFD processing limited to a balance between 300 hours and TeraFLOPS.
• Performance – Racecars weigh 100 kilos less than a Smart Car, but get heavier as they speed up, making them a bit like an upside-down aeroplane. The team make around 30k design changes to the car each year with the aim of gaining 1 second improvement per lap on average by the end of the season.
• Reliability – the best performance is of no value if your car doesn’t finish (as the failure of Lewis Hamilton’s Mercedes at Malaysia demonstrated); a team needs to design the c.100k components of each car to withstand the rigours of full speed racing.

While a Formula 1 team’s overall aim each year is to win each race, the overall driver and team championship, it’s activities are not widely different from automotive and aeroplane manufacturers, or technical research organisations.

Red Bull Racing’s applications

The team uses IT for a number of specific applications, including:

• Computer Aided Design (CAD) – the team uses Siemens for CAD and PLM.
• Computational Fluid Dynamics (CFD) – these are based on ANLYS software. Applied to Finite Element Analysis, CFD is useful in modifying/reducing the weight of components.
• Monte Carlo simulations – used along side CFD and telemetry data for ‘what if’ decision making.
• Driver simulation – each track can be driven in virtual reality, allowing new corners to be learnt before the race itself.
• Wind tunnel simulation – used in balance to validate the modification results of the CFD process
• Full car testing – each race practice day adds to the 8 days of pre-season testing each year.

The sport’s regulators limit the time and processing applied to all of these apart from CAD, Monte Carlo and driver simulation, making efficiency extremely important to Red Bull Racing’s computer usage. Red Bull Racing uses ‘heterogeneous’ processors and has made great use of AMD’s Phenom x86 chips. In fact on our tour of the factory we saw a computer room full of IBM processor racks – relatively old, but still of great use for cluster computing. The team uses MPLS connections between the factory and racecourses.

Red Bull Racing uses IBM Spectrum Computing software and HPC hardware

IBM brought Platform Computing in January 2012 and with it, its expertise and software for managing compute- (and data-) intensive applications such as simulations, computer modelling and analytics in distributed technical environments. Last year IBM created the ‘IBM Spectrum Computing’ brand for this software in a similar way to its IBM Spectrum Storage and both are Software Defined Infrastructure (SDI) divisions of the Systems group. IBM Spectrum Computing offerings include:

• IBM Spectrum LFS – workload and batch scheduler for use across distributed, heterogeneous compute systems and HPC clouds.
• IBM Spectrum Conductor – integrates IBM Spectrum Computing resource management with IBM Spectrum Storage data management and protection elements; the ‘with Spark’ version is designed for users of Spark in-memory analytics applications.
• IBM Spectrum Symphony – policy-driven resource management for High performance analytics optimises IT resources on-premises and in the cloud.

Rather than addressing the virtualisation and management of traditional commercial workloads which VMware dominates, IBM’s focus (and Platform Computing’s before acquisition) is on the High Performance Computing (HPC) environment which uses scale-out clusters of processors, (typically) Linux operating systems and Open Source virtualisation and systems management. Red Bull Racing is a typical HPC user in its design and engineering focus and need for advanced Computational Fluid Dynamics (CFD) applications. HPC architectures are becoming increasingly important for general purpose use in commercial organisations, especially for analytics (such as Hadoop Big Data Analytics, Spark in-memory and MongoDB databases) and will in future be needed for new cognitive, IoT and machine learning applications.

IBM Software Defined Infrastructure (SDI) software offerings

Source: ITCandor, 2016

In addition IBM’s HPC and SoftLayer customers can take advantage of its Aspera fast data transport software, which is superior to the use of simple TCP-IP.

The winning results

I’ve visited a number of Formula 1 teams over the years including Ferrari, Williams, Lotus and Jaguar. They all shared the same basic aims of being viable and winning races and they are all high technology engineering firms. My Red Bull Racing visit was fascinating (and very unusual) in being in the middle of a race weekend and – as it turns out – very timely as the team came in first and second. While its immediate success was due mainly to the failure of Lewis Hamilton’s Mercedes (illustrating the need for reliability), its medium and long-term prospects are based on deep technological work, which gives us an insight not just into the future of the automotive industry, but also of enterprise computing. If you’re in a commercial organisation beginning to adopt HPC-based (and often Open Source) applications and/or cluster computing you could do far worse than following Red Bull Racing as a user and IBM as a supplier to keep up with potential new directions.