Guest blog: Competing to bring a driverless future to life

Graham Cousens

Graham Cousens, autonomous vehicle business lead at Quantum, discusses the importance of data management for car makers.

The world’s leading automakers and community of suppliers, technologists and engineers are competing to bring a driverless future to life, as the connected vehicles (CVs), autonomous vehicles (AVs) and adas economy is predicted to grow into the trillions by 2050.

The goal of this technology development is to create vehicles that can drive as well or better than humans, thereby reducing traffic-related fatalities, allowing for cost-reductions and efficiencies in mass transportation and shipping, and allowing people to enjoy the ride rather than having to focus on driving.

To develop this level of technological sophistication, companies are arming test fleets of vehicles equipped with advanced GPS, lidar, radar and other sensors that are gathering massive amounts of unstructured data. Those data are then being made available to a wide array of analysts and technicians who carry out the testing required to develop advanced driving algorithms – the brain of the cars of the future – before sending the data off for long-term storage.

 

Biggest data ever

The amount of data currently being collected, protected, analysed and stored is well beyond what most industry prognosticators ever envisioned. In 2017, Intel predicted an average collection rate of 4Tbyte per vehicle, per day. In 2019, that number has exploded to 20Tbyte per hour, which equates to 50Pbyte per year.

The existing infrastructures among suppliers, service providers and partners are bursting at the seams, with nearly every company struggling to address the challenges that come with having to manage such huge amounts of data. The problem is not limited to capacity – with these challenges comes a need for speed, in having to make huge swaths of data available to geo-distributed HIL farms, including overseas, quickly and securely.

Of course, once a company sets out to address the challenges of capacity and speed, it becomes a matter of cost. Most are finding this to be much more complicated than simply doubling or tripling the existing infrastructure. This approach becomes far too costly and unwieldy, especially with the increased time and cost associated with retrieving data from a cold storage tier.

 

Golden rule

As data repositories grow in size and complexity, so does the ability to keep those data secure. Highly complex IT infrastructures often consist of siloed structures that hinder efficient data management and data security.

The key to intelligent data protection is the classic 3-2-1 rule, where at least three copies of data are stored on two different media, while one copy is held off-site. New concerns around ransomware suggest one of those copies should also be stored offline. Even in the modern mobile and multi-cloud era, the 3-2-1 rule is still the best way to protect data against hardware failures, cyber-attacks, data corruption and data loss.

 

Automated multi-tiered storage

Challenges associated with capacity, performance and cost may be addressed through a multi-tiered data management strategy. While speed and performance is required to carry out the high performance computing and analysis required to develop artificial intelligence, such high performance storage isn’t required for long-term data archiving and retention. Tape and public cloud technology remains the best option where low cost, high security and flexibility are the primary concern.

A multi-tiered storage strategy based on the policy-driven transfer of data to the media tier that achieves the optimal balance of time, security, performance and cost – without impacting user experience – ultimately results in reduced costs and improved productivity across the enterprise.

Automated technology – commonly referred to as active archives – lets end-users locate and launch their files without knowing where it physically resides, which is critical in CV, AV and adas development environments that rely on collaboration. As long as data are always placed on the most suitable media, automotive businesses will be able to manage both their productivity and their costs.

 

Summary

To compete, today’s auto manufacturers and their technology partners must recognise the importance of large-scale, intelligent data management to further the development of our driverless technology. This intelligent multi-tiered technology exists, and those companies who fully embrace it will empower themselves to be successful in the driverless economy of the future.

 

Graham Cousens is autonomous vehicle business lead at Quantum

 

www.quantum.com

 

Recommend to Colleagues: