TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
The Pragmatic Engineer published excerpts from Sam Newman’s 2026 book, Building Resilient Distributed Systems, covering disaster recovery and resilience for AI-powered software. The material discusses recovery targets, failover approaches, provider and model changes, observability, and planning for infrastructure constraints; it does not present new incident data or prove that AI coding tools have increased outages.
The Pragmatic Engineer has published excerpts from Sam Newman’s 2026 book, Building Resilient Distributed Systems, outlining how software teams can prepare for major outages and account for AI-related dependencies. The material focuses on practical planning, including recovery objectives, failover options, model-provider resilience and observability, rather than reporting a new outage or a measured change in reliability.
The excerpt’s disaster-recovery chapter describes failures that can take out an entire site, such as fires, floods, electrical problems or faults in underlying cloud infrastructure. Newman frames disaster recovery as a set of procedures for restoring service after such events, while emphasizing a trade-off: reducing acceptable downtime or data loss generally increases cost and complexity.
Two targets help teams make those trade-offs explicit. The recovery point objective (RPO) sets the maximum tolerable age of recovered data, and therefore the potential data loss. The recovery time objective (RTO) sets how long a system can remain unavailable. Reaching an RTO can involve restoring data, making computing capacity available and redirecting traffic—not simply restarting an application.
The Pragmatic Engineer’s overview says the book also covers backup-and-restore, pilot-light, warm-failover and active-active approaches. Its AI-focused topics include dependence on model providers, changes to models over time, observability, using multiple models or vendors, making costs visible, and regularly reviewing vendors. The source describes these as subjects addressed in the book, not as guarantees that any particular design will prevent outages.
Recovery Targets Shape Resilience Plans
The guidance matters because teams cannot plan for recovery until they specify how much downtime and data loss their service can tolerate. Those limits affect architecture, staffing, testing and spending. A service with a short RTO may need alternate capacity ready to take traffic; a less demanding target may make a slower, lower-cost restoration plan acceptable.
AI-powered systems add dependencies that may change independently of a software team’s own releases. A model provider can alter availability, pricing or model behavior, while access to suitable computing resources may be constrained. The article’s overview points to multi-model and multi-vendor planning and observability as ways to manage uncertainty. They also introduce costs and operational complexity, so they are choices to evaluate against a service’s needs, not automatic requirements for every product.
The source raises the prospect that AI coding agents may contribute to more code changes and production bugs, but it supplies no outage statistics or causal analysis. That claim should be treated as the article’s framing, not established evidence that AI tools have increased incident rates.
disaster recovery software for distributed systems
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Site Failures to AI Dependencies
Newman’s book addresses both failures within individual service instances and larger events that can disable substantial parts of a system. For the latter, it uses “site” as a general term that can refer to a cloud region, data center or comparable infrastructure location. The same broad recovery patterns can apply across public-cloud and on-premises settings, though implementation details differ.
The Pragmatic Engineer says Newman spent three years writing the book, one year longer than initially planned, and revisited its material to account for AI. The publication presents excerpts from two chapters and says the book has 16 chapters. It identifies the book as published by O’Reilly Media and says the excerpts are reproduced with permission.
The source is an editorial publication and book excerpt, not an independent study of system failures. Its value here is to present Newman’s framework and topics for engineering teams; readers should distinguish that guidance from measured evidence about how often outages happen or what causes them.
“The recovery point objective (RPO) defines the maximum amount of time that can have passed since the data was last backed up or synchronized.”
— Sam Newman, in the excerpt from Building Resilient Distributed Systems
As an affiliate, we earn on qualifying purchases.
Outage Trends Remain Unquantified
The material does not provide data showing that AI coding agents have increased outages, nor does it establish how much model changes or infrastructure shortages affect incident rates. Those points appear as concerns and planning considerations in the publication’s introduction, not as quantified findings.
The excerpts supplied also do not give specific RPO or RTO recommendations for particular types of services, nor do they compare the cost or performance of each failover pattern. The appropriate target and architecture depend on a system’s requirements. It is not clear from the source how the book evaluates those trade-offs in full or whether its later chapters offer additional detail on multi-vendor recovery.
As an affiliate, we earn on qualifying purchases.
Teams Must Set Their Own Targets
The next step for an engineering team applying the guidance is to define acceptable data loss and downtime, then test whether its chosen recovery approach can meet those targets. That work includes checking that backup or synchronization intervals support the RPO and that replacement capacity, data restoration and traffic redirection fit within the RTO.
For AI-dependent services, teams can also map which providers and models are critical, monitor service behavior and costs, and review vendor dependencies as they change. The source does not announce a new standard, regulatory deadline or forthcoming incident report. It points readers to Newman’s book for further treatment of these issues.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the news in this report?
The Pragmatic Engineer published excerpts from Sam Newman’s 2026 book on disaster recovery and resilience in distributed and AI-powered systems.
What do RPO and RTO mean?
RPO describes the maximum acceptable gap between the latest backed-up or synchronized data and an incident, indicating potential data loss. RTO describes the acceptable time to restore service.
Does the source prove AI coding tools are causing more outages?
No. The introduction raises that concern but supplies no outage statistics or causal evidence establishing that AI coding tools have increased production incidents.
Which disaster-recovery approaches does the report mention?
The overview lists backup-and-restore, pilot-light failover, warm failover and multisite active-active as approaches covered by the book.
What remains uncertain about AI system resilience?
The source does not quantify how model changes, vendor dependence or CPU and GPU shortages affect outages, and it does not prescribe one recovery target or architecture for all systems. Those choices depend on a service’s needs and constraints.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
![Rails World 2026 Opening Keynote [Video]](https://qatrial.com/wp-content/uploads/2026/09/rails-world-2026-opening-keynote-video-featured-560x300.jpg)
