The Datacenter as a Computer

Warehouse-scale computers (WSCs) power cloud computing and all the great web services we use daily, including recent advancements in Artificial Intelligence/Machine Learning (AI/ML). This book examines how WSCs treat the datacenter itself as one massive computer designed at warehouse scale, with hardware and software working in concert. The book details the architecture of WSCs and covers the main factors influencing their design, operation, and cost structure, as well as the characteristics of their software base. Each chapter contains real-world examples, including detailed case studies and previously unpublished details of the infrastructure used to power Google’s online services. Though targeted at the architects and programmers of today’s WSCs, this book will also be broadly interesting to those who just want to understand the infrastructure powering cloud computing and hyperscale and AI services.

The fourth edition reflects six years of advancements since the previous edition and includes significantly revamped and enhanced content including four new chapters, new case studies, and completely-redone illustrations optimized for online reading. New topics include exascale AI processing, software-defined infrastructure, advanced networking, modular datacenter design, chip packaging and interconnects, silent data corruption, trusted computing, sustainability, and using AI to design better systems. Also new to this edition, the book includes a retrospective discussing key innovations and learnings from 25 years of warehouse-scale computing, and an annotated bibliography of our selection of top 25 papers from this period. Updated discussions of emerging trends and opportunities ensure that this revised edition will remain an essential resource for educators and professionals working on the next generation of WSCs.

This is an open access book.