Platform Engineering

Concluding Remarks

A question we asked ourselves before writing this book was whether platform engineering is just a fad, like many others¹ that have swept the industry over the last 25 years. And while we hope that the ~100,000 words we’ve written on the subject have convinced you otherwise, we still see a real risk of platform engineering falling victim to the tech hype cycle. With vendors pushing specific implementation details and consultants writing checklists, we’re in danger of losing all nuance in favor of some metrics for an executive’s dashboard.

That’s why we spent so much time in Part I talking about the inflection point of complexity that the industry is facing today. While the cloud, vendor, and OSS ecosystems have accelerated technical innovation where it’s demanded by the business (in 2024, it’s anything related to AI), they haven’t made software systems any simpler than they were in the ’90s with servers as pets and proprietary vendor platforms. The frustration has shifted from dealing with data center engineers to grappling with the copy-and-paste nature of Terraform “codebases,” but the complexity remains. In fact, with the continual push for innovation and the need to integrate legacy systems, the overall complexity has grown, leading to the over-generalized swamp we discussed back in Chapter 1.

The industry needs to move past seeing platforms as either hype cycle implementation details (such as IDPs²) or the ad hoc glue that today’s hot application team uses ...