Mercedes and Pure: Optimizing off track performance with storage for a virtualized environment

At the risk of cliché, it is still a truism that performance on a Formula 1 track can’t be attributed to just the car. While the car arguably has the biggest impact on results followed by the driver, there are systems, people and infrastructure that all need to work together over an extended period of time to allow the car and driver to do what they do best. Likewise, the performance of virtualized environments can’t just be attributed to virtual machines. The underlying infrastructure and the network all play a role, but the storage layer is the piece that can make or break it.

The Impact of Latency on VDI

In a previous post, we discussed how Mercedes-AMG needed high performance and security for their sensitive trackside data - and starting in 2014, All-Flash became integral to Mercedes-AMG’s Formula 1 success.

This shift to Pure Storage also allowed Mercedes-AMG to implement a VDI environment that provided the necessary security for the sensitive, competitive performance data being generated and managed trackside. These VDI environments have tunnels directly to trackside SQL databases running on Pure Storage arrays, as well as secure portals back to the factory’s (at the time) legacy equipment.

What quickly became apparent, was that when they were connected to the local Pure Storage arrays the experience was great, but when connecting to the legacy arrays in the factory, it was the complete opposite. Latency issues resulted in wasted time for incredibly expensive on-track resources. This performance deficit wasn’t a function of the communications technology, or bandwidth back to the factory or the VMware environment – it was fundamentally a problem with the legacy storage arrays and the way they were configured, and Pure knew we could help the team.

Next Steps

As a proof point, Pure Storage setup a small factory-based array for the central IT team so they could experience the performance and management differences with the Purity OS running on a FlashArray. The difference was night and day with the VDI experience becoming seamless - whether connecting to on-track arrays or factory arrays - and it was from that point the entire Mercedes-AMG team started to experience the ‘Pure’ way of doing things. The technical team were spruiking the better environment on one side – and then on the other was the expansion and replication of the solution across the organisation.

The Pure Storage solution for the off-track operations saw a consolidation of applications, simplified integration and real-time data streams from the track, through to the factory, the office and beyond. By using Pure Storage, Mercedes-AMG were able to eliminate a lot of the frustrating lags that the data teams were working with – whilst also increasing productivity in all aspects of their F1 business.

In 2017, the conversation then turned to, ‘What can we do with the HPC grid?

High Performance Computing (HPC) uses parallel processing and supercomputers to run complex applications. It is used to solve difficult technical problems quickly and efficiently by combining three key components: compute, network and storage. With Mercedes-AMG using multiple complex simulations, data analysis and complex design programs, optimising the performance of their HPC was critical to performance both on and off the track.

Inside the factory, the legacy systems were replaced with Pure’s FlashBlade storage to handle high-performance workloads, drive productivity and protect data. Most importantly the FlashBlade was able to provide a significant performance increase where the supercomputer’s CPU and GPU utilisation reached 100%. In the world of F1 this was a huge leap forward as it allowed the engineering, data science and CAD teams to iterate more quickly.

In today's environment, there is virtualization of the majority of Mercedes-AMG server infrastructure, and VDI is used across the board. However, some of the more technical applications use the traditional paradigm of application, operating system and infrastructure – rather than using multiple applications, VMware layer, Multiple OSs. But irrespective of whether there's virtualization sitting above it or not, Pure Storage is there, assisting, supporting and optimising both environments.

In terms of backup and recovery, Pure FlashBlade enables quick and easy data recovery with low recovery-point objective (RPO) in the event of a disaster. It provides snapshot-based file system replication, supporting the NFS environment and as the backup target for the VMware environment.

The result? An active-active data centre.

For Mercedes-AMG, their investment in this highly automated, active-active environment paid off when one of their factory data centres was inundated with flood water after the riverbanks broke. Pure were able to move important workloads and data into another facility and continue with minimal disruption to their operations.

The strategy at Mercedes-AMG wasn’t about compressing the amount of VMs, or making a particular metric look more efficient. It was about speeding up and improving the VMware environment itself, resulting in:

  • Better backups
  • Simpler processes
  • Simpler operations
  • Optimised databases

Pure Performance

Today Pure Storage is ubiquitous across the Mercedes-AMG environment, with all major parts of the Formula One team strongly involved with Pure Storage. By virtualizing their HPC workloads and underpinning the environment with Pure’s solid state storage, Mercedes-AMG have added flexibility, efficiency and security whilst scaling their overall capability and absolutely improving performance in all aspects of their organisation.