BMW Logo
Sector: Automotive
Ubicación: Germany

BMW: Creating a continuum between cloud and edge for enhanced data management and quality assurance

Aspectos destacados

  • By digitizing its factories, BMW has continuous data management over the full production lifecycle.
  • Being data driven in its factories allows BMW to save time and energy.
  • BMW has improved the quality assurance process using high-definition cameras and AI to detect problems.
  • BMW creates and deletes 2.6 million VMs per month.

Productos

At-a-Glance

In June 2023, SUSE’s Ivo Totev and Dr. Thomas Di Giacomo spoke with Andreas Pöschl, senior solutions architect at BMW AG, about how the vehicle and motorcycle manufacturer innovates using data and AI, as well as how the company sees the role of the edge in cloud computing. 

Watch the interview

“It's not about using VMs in public cloud, it's about using the technical platforms, using the functional platforms, and creating this continuum between cloud and edge. It's not cloud, and edge is somewhere else. Cloud is also not somebody else's computer. It's a usage and operation model. You need to change the mindset for that.” 

Interview transcript

Yesterday, I drove down to Munich from Heidelberg, which is 350 kilometers away. I drove an electric BMW, where SUSE is inside, and the operating system powers many functions of that vehicle. I couldn't be more happy now that we have with us the head of edge computing and container runtime at BMW. So welcome, Andreas Pöschl. 

You are a great community member. You're a great supporter of SUSE for many years, and we have worked together to drive a transformation at BMW. Maybe you can share, to start with, some of the stages we have gone through together to transform how BMW operates today. 

Our journey together in this open source, business-critical environment started 2007/2008, where we struggled together to get fiber channel, host-based adapters, kernel drivers, up and running in SLES 9 SP3. We started to do it for database workloads, which was a big move from UNIX systems to this small open source Linux system. Briefly after that, we started to do virtualization, and we also did it together with Xsens. We worked a lot to get it up and running and to fix all the limits we found. We always tried to stay open there, and that perfectly matched with your approach of the ‘open, open.’ It was not only about open source. It was about this open mindset being exchangeable. We wanted to be able to change, if necessary. We wanted to avoid lock-in. When we moved on to OpenStack, we loved OpenStack and Ceph. There, we could already use this ‘open, open,’ because we had to move to a different distro. 

Briefly after, we have been the first customer of Harvester. We are now using Harvester and Rancher in this edge ecosystem. It still is about this cooperation, this togetherness. We are not personal member of the community, but together we give you the use cases, the test cases, and I think we can do great work for the whole community together. 

I want to ask the same question: in terms of what's the next big innovation top of mind for you, where do you place your bets? 

Innovation is not about fancy ideas. Actually, innovation is if you have got a business benefit. Otherwise, it's just weird ideas. Leonardo da Vinci, one of the most innovative people who ever lived, did war machines, and helicopter and parachute-like things. The war machines were the innovation because that was business. It brought a benefit. The other things were just fancy ideas. 

It's the same with the enterprises and the companies out there. We need something where we benefit from it. It sometimes takes some time until we get from the ideas through the real innovation. What we currently try to do is break up all the silos regarding the data. We have got a lot of sources of data, and it’s not only the new design data, it’s also existing information about factories, for example. We are digitizing the factories, and we bring that together so we have a continuous data management all over the full process. We are able to use the data from design, through production to sales. We create digital twins of factories, so we can already simulate and run the full production process without building anything. If something changes in design, immediately you know what to change in the production line. 

That's really exciting. 

It's a really big benefit because it [saves] a lot of time and avoids a lot of waste of energy.  

Another thing is AI. Everybody's dealing with AI. I'm not talking about generative AI, now. Everybody uses ChatGPT and talks about that at the moment, but the AI we are talking about is helping to improve quality. 

We take a lot of pictures and videos in production, so we have got a lot of high density cameras around a car in the paint shop. We see if there are any issues. There are high-definition cameras to see if the bolts are really tight, if we forgot washers, if all the pieces are assembled. At the moment, we are working on automated driving in the plants. At the end of the production line, usually a worker drives a car off the production line over a small bumpy road to hear if there are any noises. What we do now is based on AI – we recognize the noises via AI; we compare it to the base spectrum. You cannot have a person in the car, so we have to drive the car in automated way. We are moving this car without any driver off the production line. 

Doing quality assurance. 

It's quality assurance, and it gives us a big improvement because AI is better in recognizing that there is something wrong. Not yet in detecting what, exactly, but knowing that there is something. For all that, we need to move to public cloud. We have got this private cloud environment based on OpenStack for build environments: 110,000 builds per day, 2.6 million VMs per month created and deleted. Still, we need to move up in the layers. 

2.6 million VMs per month created and deleted. You fight for every second there, and you need to move up in the platform. It's not about using VMs in public cloud, it's about using the technical platforms, using the functional platforms, and creating this continuum between cloud and edge. It's not cloud, and edge is somewhere else. Cloud is also not somebody else's computer. It's a usage and operation model. You need to change the mindset for that. If people don't understand that there is something bigger there than just using a lot of VMs, you will not be able to leverage the benefits of public cloud. 

You mentioned two things. The data example was about breaking silos for data, and here, it's breaking silos for compute in a way. So, having a continuum of data and compute across different infrastructure and where the business happens.  

Exactly. 

I really like that you talk about the edge, which is the next big thing, because computing will happen based on smaller units. Where the action is. Where the data occurs. You obviously already do a great job in that. Coming back to our theme, though, about the offense and defense – you also have to play that, I assume. Where do you get really innovative, and where does it have to be rock solid so that you can sleep well? If you could share a little bit on how you think about it. 

It's actually about awareness. It's nice to see the innovation, but don't forget about resilience. A lot of people, especially developers, would like to move on to innovation, but then they complain that things break now. It's moving together, business and IT. IT is business, and business is IT. It's not separate silos anymore. You bring together things like Dev and Ops; it became DevOps. We forgot the business. Biz DevOps. What about security? BizSecDevOps. Now it costs a lot of money. BizSecFinDevOps. 

With this awareness, think big, and create this continuum. Talk to each other. It's about communication. Understand that you still need those people who ran the environment for ages already. They have so much experience. Don't forget about them. Leverage them. Let them participate in this movement, in this innovation. We try to let the same teams who ran something in the past infrastructure move on to the new infrastructure because then they are not forgotten about. They can bring in their experience and their knowledge into the new technologies. I think this is crucial. 

Well, that was really nice. BizSecFinDevOps. 

Let's see what we talk about in some years, right? 

Andy, we're on time. I cannot believe you're on time actually, but you're on time. I want to see you in another of 15 years for the next SUSECON. Thank you very much for coming.