On the road again

The best or nothing

As the inventor of the automobile, Mercedes-Benz, a brand of the Daimler Group, has pushed the limits of the concept of an automobile, from introducing the first multi-valve engine to setting new standards in safety testing and performance engineering. But constantly testing the limits of automotive technology takes more than just imagination and a love of driving.

It requires being able to combine modern computing technologies with advanced data modeling capabilities to achieve new innovations and breakthroughs. And in the age of connected devices and autonomous vehicles, it requires building a robust and agile cloud infrastructure capable of connecting and coordinating the billions of data points that will form the 'brain' that drives its connectivity, a job well suited for Mercedes-Benz Research and Development North America.

"Daimler created a research and development hub specifically to build the foundation for its connected and autonomous vehicle portfolio under the Mercedes brand," Sriram Govindarajan, principal infrastructure engineer at Mercedes-Benz Research & Development (MBRDNA) says. "Almost immediately, we recognized that in order to build products of tomorrow we had to first build an IT infrastructure suitable for today, which meant moving the majority of our existing infrastructure to the cloud and modernizing our development processes at the same time."

Zero visibility stifles performance

Connected cars have raced to the forefront of the Internet of Things (IoT) trend, designed to share data in real-time with other systems outside the vehicle to dramatically improve safety, performance, and passenger comfort. Each aspect of that connectivity is powered by a purpose-built app, and each app is out powered by an array of microservices that must connect with other services to complete the system.

With Microsoft Azure as its cloud platform, MBRDNA adopted Kubernetes to containerize development and provide the speed and agility required to deliver new features faster. "Kubernetes was ideal for us because it centralizes the code base, infrastructure, and everything else our teams need to quickly develop the features and capabilities in our roadmap," Govindarajan says. "But the more services we ran in each cluster, the harder it became to match them up and connect them because we didn’t have the right service discovery mechanisms in place."

Govindarajan says that development teams may have services running in their individual clusters and had no way of identifying other services or dependent services running on other clusters and given how quickly service addresses, hosts, and ports are added and changed, doing so without a dedicated service discovery tool was virtually impossible. "Service discovery is 100 percent essential to our entire operation and to the ability to meet our responsibilities to the broader organization," he says. "A developer could scour our entire environment for days and never find what he or she was looking for, so we needed a way to accelerate discovery and do it effortlessly across a number of clusters and clouds."

Govindarajan says that his team's success to this point inspires confidence for how they'll use it in the future. "Consul proved to be the right solution for solving our service discovery challenges that we found at the right time," he says. "We're eager to start using the upgraded features in the next versions of Consul to continue strengthening the resilience and performance of our infrastructure and help usher in the next generation of innovative driving machines."

In addition to having realized value from Consul immediately as a unified Service Discovery tool, MBRDNA plans to adopt Consul in the future as a service mesh to deliver service-to-service communication for their distributed applications. This includes enabling application teams to build out more flexible deployment scenarios such as Canary and blue/green testing.