Site Reliability Engineering, 2nd Edition

Chapter 20. How SRE Supports AI

By Dan Dennison, Isaac Albarrán Muñoz, John Lunney, and Rita Lu with Garrett Holthaus

This chapter is one of two in this book that discuss the intersection of AI and Site Reliability Engineering. Here, we focus on the application of the various aspects of the SRE discipline to the operation of AI services. Chapter 21 concentrates on the application of AI to the SRE discipline.

Artificial intelligence is the overarching field dedicated to creating machines capable of human-like reasoning and problem solving, encompassing everything from simple expert systems to autonomous agents that use large language models (LLMs) as reasoning engines. While machine learning (ML) is a specific subset of AI that allows systems to learn from data patterns without explicit programming—powering tasks such as image recognition and fraud detection—the two fields often overlap. This is most evident in technologies like LLMs, which use deep learning (an ML technique) to achieve the sophisticated AI capability of understanding and generating human-like language, effectively blurring the lines between data-driven learning and broader intelligent behavior.

The wide proliferation of artificial intelligence and machine learning applications has not left the discipline of SRE untouched. In fact, the first Google SRE team supporting ML applications was founded in 2009 to support the Smart Ad Selection System,¹ which had been created five years earlier to supply the Google Search ...