Will Platform Engineering Get Loved To Death?

The term platform engineering is enjoying a surge in popularity.

Yet the value of the term risks collapsing under the weight of its own popularity, hugged to death by over-eager marketing folk keen to jump onto the latest trend. The risk was apparent to me during a panel session on Platform Engineering at KubeCon EU on Thursday this week.

“Platform Engineering is really about giving the piece of infrastructure that you need to your teams, but more as a product,” said Sarah Polan, Field CTO for EMEA at HashiCorp. “For the past couple of years, I think the previous trend was very much DevOps.”

Indeed the DevOps trend has been so strong that Ops became the go-to suffix, applied to almost any other term as marketers competed to define a new category all for themselves. SecOps, DevSecOps, GitOps, FinOps, MLOps. Anything to try to make whatever was already happening seem new and different and fresh.

Of course, this behaviour is nothing new; we saw it play out with previous hot trends. Agile, cloud, SRE, all these terms were eagerly applied to situations with nebulous connections to the original intent at best. Instead of helping customers understand how to solve their actual problems, new labels were slapped on old things. A magic bullet will definitely work, this time for sure.

“DevOps isn’t something you buy, but many companies have tried to sell it to you,” said Stu Miniman, Director of Market Insights and Hybrid Platforms at Red Hat. Many of them succeeded, and plenty of invoices were paid.

Those deeply involved in the industry are sure they understand what platform engineering is, but it’s not clear to me that they agree with each other. Nor is it clear that customers understand the term. With 58% of attendees at KubeCon attending for the first time, there are many customers that are still learning what cloud native is about. It’s unclear to me if they are able to quickly arrive at a clear understanding of what platform engineering actually means.

“Platform and product can literally be the same thing, depending on what you’re talking about,” said David DeSanto, GitLab Chief Product Officer at GitLab. “For us at GitLab, the platform is the product.”

This drive to call any product a platform has been prevalent across the industry for some time. While GitLab can more credibly lay claim to be one than many, the term platform has itself been devalued by every venture-funded startup selling a feature masquerading as a product that claimed to be one. Do we really want to be engineering more of them?

The question for us as an industry is whether we are willing to let yet another useful term get so over-hyped that it loses all value and becomes yet another empty buzzword. I think platform engineering—far more than DevOps or even cloud—is too useful a term to let that happen. Not without a fight.

HashiCorp’s Polan provided what I think is a good first draft of a clear definition for what platform engineering is, and what it is not.

“For me it’s something that supports other infrastructure or business logic,” said Polan. “Anything that would be business logic or solving a business problem probably falls outside of platform engineering and would be more on the application side of things.”

I’d like to see this extended to create something that customers can use to quickly evaluate vendor claims about what they’re being sold. If it’s too difficult to determine if something is or isn’t platform engineering, or if consensus on what the term means can’t readily be found, perhaps we can cut to the chase and declare platform engineering dead already and move on to the next shiny toy.

I hear generative AI is all the rage.

I attended KubeCon+Cloud Native Con EU virtually as a guest of KubeCon.