General Motors (GM) is one of America’s most iconic companies—the Chevrolet logo, in particular, is synonymous with American muscle cars and cruising the open road—but today China is the company’s largest market. In fact, it’s responsible for 40 percent of the company’s global sales. So when a new Chinese law or regulation might significantly affect those sales, GM takes notice.
In a talk at SpringOne Platform 2019, Zac Grantham, Senior Development Manager at GM, and Tim Schneider, Senior Manager for Cloud Services at GM, discussed just such a scenario. Namely, a Chinese regulation passed in late-2017 mandating that, by the end of 2018, all data that originates in China must remain in China.
Twenty years ago, this probably would not have been a big deal at all. But today, when many GM vehicles are equipped with infotainment systems that send personalization data between car and cloud, it’s a very big deal. The main problem was that GM had outsourced some critical infotainment-system development and hosting to a third party, which had no servers in China and didn’t intend to put any there.
“It wasn't a planned thing. It immediately became a crisis for us at GM,” Grantham explained. “. . . Basically, with this law being passed, overnight, we had a ticking clock until our vehicles became illegal."
The situation in which GM found itself looked something like this:
It had just over a year to rebuild its application and underlying platform, all the way down to the infrastructure—which had to be physically located in China.
It also had to account for testing of the software in actual vehicles, a process that typically takes months and can be affected by everything from rain to dead batteries.
In the end, Grantham, Schneider, and their respective teams decided they realistically had about 4 months to develop their new applications, in order to account for testing, infrastructure, and any other obstacles that might arise in the meantime.
Prior to this, the company had been experimenting with Pivotal’s products in pockets—including within Grantham’s and Schneider’s teams—but now seemed like the time to go all-in.
“And that was a big step for us, as a company,” Grantham said. “To say something that is so critical to our company as a whole—is sort of a threat to our existence, potentially—we're going to go all in on this technology to solve it for us, was a big step for us to take.
“. . . In late-2017 when we started this, we basically went and looked at every single breakout session from the 2017 [SpringOne Platform] conference, took all the buzzwords, and said, ‘Yes, let's do that.’ If you went back two years . . . every session was [about] Spring Boot and Spring Cloud, and a lot of microservices. And so we looked at what was there and we . . . went all in.”
Obviously, GM is still around and selling cars, so the bet on Pivotal Platform was successful. Grantham and Schneider go into a lot more detail about the decision-making process, implementation, and results in the video above (which includes an informative Q&A session), but here are some more highlights from the presentation.
Moving fast and focusing on the business problem at hand
Grantham: “We had actually a lot of college hires, sort of brand new college hires, [and] this was a lot of their first experiences, and a lot of the tools were easy for them to grasp. You can work with Spring Boot really quickly to spin up a service. Spring Cloud allows you to have a circuit breaker very quickly and easily. You can use Hystrix for that. You can use the service registry. Very simple. . . .
“We didn't have to spend a lot of time setting up these microservices. Looking at boilerplate code, we could very quickly spin up some templates and get going on the actual, real business problems we were trying to tackle, not just piecing together technology.
“But that software on its own wouldn't have done anything without Pivotal Platform and the Pivotal Marketplace underneath. I mentioned we had the concept of a service registry to connect all of our microservices. Well that took 30 seconds to set up with the Pivotal Marketplace in Pivotal Platform. We use RabbitMQ. Again, it's 30 seconds to set up and bind our services. So by combining the Spring software library choices with the Pivotal Platform as our platform, we were able to marry that and just, again, start moving forward as quickly as possible.”
"We could very quickly spin up some templates and get going on the actual, real business problems we were trying to tackle, not just piecing together technology."
Automation saves the day when everything else is complex
Schneider: “Quite honestly, one of the bigger challenges was, once it was [in China], where was it going to run? So . . . what we ended up doing was working with our joint venture partner, that's SAIC. For those of you who don't know, in order to do business and to sell vehicles inside of China, you actually have to be part of a joint venture where the Chinese side of it owns 51 percent.”
Grantham: “It basically means as we talk about taking this to production, it's taking it to production at another company. Like truly another company. . . . [W]e're two separate companies trying to accomplish different things directly 12 hours apart. So it's tricky dealing with all of that.”
Schneider: “. . . We had to go into the North American cloud—we had to develop there, we had to test there, validate that everything worked—and then we were able to use the Concourse pipelines to then deploy to the newly set up platform in China.”
Schneider: “. . .After the install, we obviously talked about training. You have to remember: Once we set this up, 12 hours on the other side of the planet, we needed staff on the ground with hands on that could actually maintain and operate the system for us. Otherwise, we're going to be getting calls all hours of the day and night. So part of the plan that we had was to go in and do a lot of knowledge transfer to our [joint venture] partner so that they could actually maintain the system itself.”
Grantham: “ . . . From the application perspective, it looked exactly the same when we used our North America install versus the later install in China. So we're able . . . as a development team, to test against our North America GM-internal Pivotal Platform. And then later, once it was ready in China, immediately switch testing there with no delay. There wasn't any downtime. It was one day we're in North America, next day we're in China. Everything works seamlessly because everything was abstracted away by the Pivotal Platform.”
Schneider: “Just to summarize a lot of what we've been saying: We ended up getting a local China environment. It came in on time. It was fully tested. We had a redundant and scalable thing, with trained onsite support, and ended up making the regulatory timeline that we had to make in order to continue selling vehicles.”
"From the application perspective, it looked exactly the same when we used our North America install versus the later install in China. . . . Everything works seamlessly because everything was abstracted away by the Pivotal Platform."
Success breeds scale
Grantham: “This sort of started our snowball. . . . I have the numbers for this particular app—just the app store and personalization. It's one of our hundreds of apps now, at least. But I mentioned we started in China, and that was late last year. And since then, we've gone to 3 other regions. We're in 4 different regions around the world, and in each of those regions, sometimes up to 20 different Pivotal Platform spaces, where we have 19 different microservices as part of this application, and each of them obviously has a number of app instances. So with just this one product itself, we have 3,000 total app instances around the world.”
Learn more about multi-cloud
About the AuthorMore Content by Derrick Harris