Chapter 7. SRE Without SRE: The Spotify Case Study
Many people are surprised that Spotify does not actually have an SRE organization. We don’t have a central SRE team or even SRE-only teams, yet our ability to scale over time has been dependent on our ability to apply SRE principles in everything we do. Given this unusual setup, other companies have approached us to learn how our model (“Ops-in-Squads”) works. Some have adopted a similar model. Let us tell you a bit about how we came to this model and how it works for us so that you can see if a similar idea would work for you.
First, we need to share a little context about our engineering culture: at Spotify, we organize into small, autonomous teams. The idea is that every team owns a certain feature or user experience from front to back. In practice, this means a single engineering team consists of a cross-functional set of developers—from designer to backend developer to data scientist—working together on the various Spotify clients, backend services, and data pipelines.
To support our feature teams, we created groups centered around infrastructure. These infrastructure teams in turn also became small, cross-functional, and autonomous to provide self-service infrastructure products. Some examples of these are continuous integration, deployment, monitoring, software ...