PlatformCon：平台工程的影响_AI阅读总结 — 包阅AI

When Vilas Veeraraghavan was director of engineering at Walmart Labs, a big challenge the retailing giant grappled with was how its developers took a variety of paths to deliver their code.

“We had too many areas, too many businesses that had goals in mind for themselves,” he said during an online panel at PlatformCon 2024, the virtual event held in June. Platform engineering, he told the audience, was how the organization found alignment among all those devs working for all those entities — and sped up productivity.

To Veeraraghavan, who worked at Walmart from 2017 to 2021, the biggest benefit of the organization’s platform engineering initiative was “when we created a single way to check in code, a single way to orchestrate delivering that code. That created a massive impact in the company.”

Even IT staff who were initially skeptical got on board once they saw the results, he said. “It was the difference between waiting two days or three days, even with continuous delivery enabled, to deliver something — all the way down to 60 minutes or 30 minutes, going from idea to production in some cases.”

Veeraraghavan, now vice president of software engineering at Truckstop, a transportation technology company, has also held engineering management roles at Bill.com, Netflix and Comcast. He joined three other experts (and me, the host) for a discussion on the impact platform engineering can have not only on developers but also system administrators, executives, customers and the bottom line.

Consistency for Devs and Ops

Platform engineering means developing an internal platform and portals that all developers in an organization use to build and deliver their code, often with the help of open source tools and templates like Backstage.

Developer productivity — notoriously hard to measure — is usually touted as the key reason to start a platform engineering initiative. But consistency is another benefit, and not just for devs, said another panelist, Nathen Harvey, developer advocate and lead of the DevOps Research and Assessment (DORA) team at Google Cloud.

Referring to the story of Walmart’s transformation, Harvey notes, “one of the things that enabled is portability within the organization. So a developer could move from one part of Walmart to another part of Walmart and they would’t be starting from absolute zero, starting fresh. They could understand: this is the process that we have.”

Similar benefits, he said “can accrue to the operations team. They know that applications are always following this path, they know that they’re always packaged in a common way, or we’ve got other commonalities across applications. That could help with troubleshooting, with monitoring, with managing the scaling up and down of those applications. That consistency breeds a lot of benefits for everyone.”

The impact of having an internal developer platform on operations engineers —the ones tasked with provisioning infrastructure when self-service dev portals aren’t part of the workflow— was on the mind of another panelist, Steve Pereira, CEO of Visible, a value stream consulting company.

Tech has been slow to understand economies of scale, Pereira said.

“You wouldn’t have an accountant in every single team in your organization,” he said. “And if you had to do accounting multiple times a day in order to get your work done, you also wouldn’t have two accountants who everybody is bugging constantly.”

In tech, the impulse is to reinvent the wheel, Pereira said. “It’s taken us a long time to realize that platform engineering is just something that we should have always had. It’s taken a while for the tooling to evolve and an understanding that this is really another area of the business that just needs to be focused on flow and streamlining operations and leveraging economic principles.”

In short, he added, “It’s about building the ability to centralize some of your best talent but protect them from being constantly interrupted, by having them develop something that can be consumed on demand and self-service.”

Veeraraghavan chipped in that Pereira’s comment about tech’s impulse to create new tools, whether needed or not, made him imagine “you want to hit a nail into the wall and every single time, the team is building a new hammer just for their company.” Platform engineering, he said, helps save resources that are being spent on unnecessary repetitive toil.

Faster Onboarding, Better Retention

Rapid onboarding of developers is another way that platform engineering has an impact on an organization, Veeraraghavan said. At a previous employer, Bill.com, he said, “The team worked on improving the onboarding of new developers. The specific focus for us was, can we get a developer ready to check in code as soon as the first week that they are a part of a team?

“We had a timeline of about six or seven weeks on average for even a staff-level engineer who was coming in to get onboarded and productive — and that’s not unlike many other companies I’ve worked for in the past,” he said.

“And so we shortened that over time by making sure that we had a specific template of tools that are available for backend engineers and frontend engineers, right in the laptop the first time they get it. And then we obviously streamlined our build process, all of that stuff.”

The result: new devs at Bill.com were able to check in their first lines of code in less than a day.

“Think about the lead time for onboarding a developer,” Pereira added. “Folks might think that that only applies when you hire someone new but it actually applies to every time you want to run a new experiment,” or anytime your devs and ops engineers need to change context, such as when building a new product or adopting new technology.

Making it easier to do so, Pereira said, also helps recruit and retain tech talent — developers and engineers will be happier, and more inclined to recommend your organization to their friends.

“If you want a real advantage, I don’t think it’s about trying to squeeze the last 10% of someone’s time out of their day,” he said. “It really is about unblocking them and enabling them to be the best version of themselves.”

In addition to helping to remove blockers that slow down code building and delivery — and ultimately, time to market — the human impact of platform engineering can be monumental, the panelists said.

Mallory Haigh, director of customer success at Humanitec, told the panel that she recently had what she called one of her favorite work-related conversations ever with a customer.

“He came to me and he said, ‘Mallory, this is the first time in four years I’ve been able to go to my kid’s football game and not bring my machine and not sit on-call and not have to worry about deployment failures or infrastructure failures. I could just focus on my kid and his game.’”

“And that, for me, was why we do this,” Haigh said. “That was a really, really wonderful story and I think I think those are the stories that come out of so many people who do platform engineering right.”

Check out the full video from PlatformCon to learn more about what the panelists think of how to measure the impact of platform engineering.

VIDEO