Michael Nygard - DeveloPassion

# Michael Nygard Michael Nygard is a software architect and author. He wrote [[Release It! (Book)|Release It! Design and Deploy Production-Ready Software]] (Pragmatic Bookshelf), the book that popularized stability patterns such as the Circuit Breaker and Bulkheads. He also wrote "Documenting Architecture Decisions" (15 November 2011), the short blog post that introduced [[Architecture Decision Records]]. Release It! came out of operations work. After a consulting company he had co-founded in 2001 wound down, he took a contract as an "application administrator" on a commerce site launch: no root access, but in charge of the web and app server configs, and later of the production infrastructure itself. He was a developer doing ops years before anyone called it [[DevOps]]. At 3 a.m. he read the stack traces and kept finding the same bugs: connections taken from a pool with no timeout, exceptions that never gave the connection back, threads waiting forever. The standard fix was a restart, several times a day. That bothered him enough to write a book about it. The first edition appeared in 2007. The second edition (January 2018, 376 pages) covers cloud, containers and continuous delivery too. The book first catalogs how systems break (integration points, chain reactions, [[Cascading Failures|cascading failures]], blocked threads, slow responses, unbounded result sets) and then the patterns that contain the damage: - **[[Timeouts (software)|Timeouts]]**: never wait forever on a remote call - **[[Circuit Breaker]]**: wrap a remote call and count failures; once they cross a threshold, the breaker trips and later calls fail immediately instead of piling up behind a dead dependency. [[Martin Fowler]]'s write-up of the pattern credits Nygard for making it popular - **[[Bulkheads]]**: split resources (thread pools, connection pools, server instances) so one failing part can't sink the whole ship - **[[Fail Fast]]**, Steady State, Let It Crash, Shed Load ([[Load Shedding]]), Create Back Pressure ([[Backpressure]]) and a few more None of these patterns is complicated. What makes the book stick is that each one answers a failure he watched happen in production, and the case studies (an exception that grounded an airline, a retail site through Thanksgiving and Black Friday) make that concrete. I list it among the books in my [[Software Architecture]] note for that reason. The ADR post is just as practical. Keep short, numbered text files in the repository, one per decision, each with its context, the decision, its status and its consequences. When a decision gets reversed, the old record stays and gets marked as superseded. It's a cheap form of [[Design Rationale]], and the format works for any project, which is why I point to it in my Knowledge Worker Kit. His roles include VP of Customer Solutions at Cognitect, the company behind Clojure and Datomic (see [[Rich Hickey]]), and a modernization program at Sabre, the airline booking system that started on American Airlines mainframes, where his mandate covered continuous delivery and moving everything except the mainframes to Google Cloud. At Nubank, conference bios have introduced him as General Manager of Data (YOW! 2023) and as Chief Architect (GOTO, 2025). He also contributed to *97 Things Every Software Architect Should Know* and still writes at michaelnygard.com. ## Recent Activity - 2026-10-08: "AI versus Vendor Lock-in", blog post (https://michaelnygard.com/blog/2026/10/ai-versus-vendor-lock-in/) - 2026-05-11: "AI versus Throughput", blog post applying the Theory of Constraints to AI-assisted development (https://michaelnygard.com/blog/2026/05/ai-versus-throughput/) - 2026-05-09: "AI Versus Microservices", blog post (https://michaelnygard.com/blog/2026/05/ai-versus-microservices/) - 2025-11-03: "Building Software That Survives", GOTO Unscripted interview with Charles Humble on Release It!, Sabre, Nubank and Conway's law (https://gotopia.tech/articles/408) ## Insights - A remote call without a timeout can block every thread that waits on it; that's how one slow dependency turns into a cascading failure (Release It!) - "Microservices were always a technical solution to an organizational problem." (AI Versus Microservices, 2026) - Conway's law says organizations produce designs that copy their *communication* structures, and most readers skip the word "communication" (GOTO Unscripted, 2025) - Every system has one bottleneck; making another station ten times faster doesn't raise overall throughput (AI versus Throughput, 2026) ## Quotes  ## Books   - [[Release It! (Book)]]  ## Media  ## Appears in  ## Related - [[Release It! (Book)]] - [[Architecture Decision Records]] - [[Design Rationale]] - [[Software Architecture]] - [[Fail Fast]] - [[DevOps]] - [[Continuous Delivery (CD)]] - [[Technical debt]] - [[Circuit Breaker]] - [[Cascading Failures]] - [[Backpressure]] - [[Chaos Engineering]] - [[Martin Fowler]] - [[Rich Hickey]] - [[Timeouts (software)]] - [[Bulkheads]] - [[Load Shedding]] ## References - Release It! Second Edition, Pragmatic Bookshelf (January 2018): https://pragprog.com/titles/mnee2/release-it-second-edition/ - Release It! first edition, first published 30 March 2007 (Goodreads): https://www.goodreads.com/book/show/1069827.Release_It_ - Michael Nygard, "Documenting Architecture Decisions" (15 November 2011): https://cognitect.com/blog/2011/11/15/documenting-architecture-decisions - Martin Fowler, "CircuitBreaker": https://martinfowler.com/bliki/CircuitBreaker.html - GOTO Unscripted, "Building Software That Survives" (2025), transcript: https://gotopia.tech/articles/408 - Data: The Land DevOps Forgot, YOW! Perth 2023: https://www.youtube.com/watch?v=459-H33is6o - Personal site: https://www.michaelnygard.com