DevOps Is the Philosophy, Platform Is the Practice

"DevOps is dead."

Well, not exactly. But the DevOps methodology of "you build it, you run it" has been failing development teams for years.

On this week's episode of Dev Interrupted, we sit down with Kaspar von Grünberg, founder and CEO of Humanitec. Listen as Kaspar explains the significant cognitive load placed on developers as a result of DevOps practices, how that has caused software engineering to be the only industry since Medieval times not to drive towards specialization, and why platform engineers provide a solution to the outdated DevOps model.

Episode Highlights:

• (2:24) What is platform engineering?
• (7:05) Should VPEs have a platform team right now?
• (11:29) Difference between SREs and platform teams
• (17:14) DevOps is dead
• (19:11) How scale affects team size
• (26:12) Standardization of the space
• (28:08) Kaspar's work at Humanitec
• (32:30) The future of platform engineering

Episode Excerpt:

Dan: If I'm a VP of engineering right now, listening to this pod, should I have a platform engineering team?

Kaspar: Yes. I mean, no, you can always argue a customer has an interest in you having a plethora of engineering teams, but I am looking at the return on investment of these teams. And there, it's so large, you can gain so much from this. There's so much inefficiency in these workflows that, yes, you definitely should have one. And having a platform engineering team doesn't like sounds, you know, more costly if you want that. It is- take a product manager, halftime if you want. But structure this correctly, structure this as a product, find a couple of people that are responsible for this, take them from different groups, you don't need me to rehire, apply these principles, you know, take this on the structure. And you'll see a very, very fast return with fairly low costs. And so definitely, definitely yes. And I want to get back to one of the absolutely correct things you said. We have these fundamentalists shouting at us. You know, everybody has to do everything in context. Otherwise, you're abstracting people away. And those are these. It's this type of thing you can always say. Everybody could always say that they say never restrict developers, never take away from this. Of course not. But that's not the idea. Like platform engineering is not about taking context away, the contrary holds true. It's about providing context. If you're looking at 700 different script formats, that's not context. That's cognitive overload. You don't win anything. And so that is really like, our industry is the only industry that is not actually driving towards specialization, from the medieval ages to now. Every industry has always specialized. We're the only fucking industry in the world that is actually working against specialization. I already have a problem with these fundamentalist approaches or viewpoints that many of them just have never really worked at scale. And scale for me means, like, production-grade two fifty-three, four or five hundred engineers over a longer period of time. And to believe that in these situations, you can just shift everything to everybody is so insanely naive. And then the next argument I always hear is like, Hey, you build it, you run it, Werner Vogels said so, I mean, let's pause and look at the situation where Werner Vogels said that he said in a blog post, 2006. That guy was in the CTO position for a UX director of research for like 12 months before he was a researcher. That guy had never worked at scale. And Amazon's teams, at that point, were a couple of dozen developers. The sentence that this guy said in 2006 says nothing about the reality of a bank in the US East with two and a half thousand developers that are drowning in policy. It's just naive to say that. It doesn't make sense.
________

Join Kaspar and I at PlatformCon 2023 on June 8th and 9th — it's virtual and free!