TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
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 emphasizes setting recovery targets, preparing for infrastructure constraints, and monitoring changing AI models and providers; it does not report new outage data or a measured increase in incidents.
The Pragmatic Engineer has published excerpts from software engineer Sam Newman’s Building Resilient Distributed Systems, a 2026 book that addresses disaster recovery and the challenges of building reliable software that uses AI. The material sets out practical recovery planning concepts and argues that AI systems also require attention to provider reliability, model changes and observability.
Newman’s excerpt on disaster recovery focuses on planning for events that disable a large part of a system, such as fires, floods, power failures or cloud infrastructure problems. It presents two measurable targets: the recovery point objective (RPO), which sets the maximum acceptable period of potential data loss, and the recovery time objective (RTO), which sets how long a service may remain unavailable. RTO includes restoring data and computing capacity as well as redirecting traffic.
The report describes several recovery approaches, including backup and restore, pilot-light and warm-failover setups, and multisite active-active systems. These options involve trade-offs: reducing acceptable downtime or data loss generally adds cost and operational complexity. The excerpt says organizations should choose targets and designs suited to their own requirements rather than assume one approach fits every system.
The report also identifies AI-specific resilience concerns covered in the book: reliance on model providers, changes to models, and the need for observability. It says the material discusses approaches such as using more than one model or vendor, making costs transparent and regularly reviewing suppliers. Newman’s book took three years to complete, according to the report, one year longer than initially planned; its author revisited sections to account for AI’s relevance.
Recovery Targets Shape System Readiness
The excerpt’s main practical point is that resilience requires decisions made before an incident: teams need to define how much data loss and downtime they can tolerate, then select and test a recovery design that meets those targets. Without agreed RPO and RTO goals, organizations may invest in costly infrastructure that does not match their actual needs—or discover during an outage that their recovery arrangements are too slow.
AI adds dependencies that can change outside a software team’s direct control. A model provider may experience availability problems, and a model may be updated in ways that affect a system’s behavior. The report’s discussion of observability and vendor planning is relevant to teams whose products rely on external AI services, though it does not quantify the scale of these risks or establish how often they cause failures.
As an affiliate, we earn on qualifying purchases.
The Book’s Disaster Recovery Framework
The Pragmatic Engineer report presents excerpts from two chapters of Newman’s book, which is published by O’Reilly Media. The excerpts span disaster recovery and resilience considerations for AI-powered software; they represent only part of a 16-chapter book. The report says the excerpts are reproduced with permission.
Newman uses “site” as a general term for a location where computing infrastructure runs, covering both public-cloud regions and data centers in private-cloud or on-premises environments. The recovery patterns discussed can apply across these settings, while the detailed implementation will depend on a system’s infrastructure and requirements.
The report’s introduction says software teams are seeing more outages and suggests that more code deployments, including code written by AI agents, and more production bugs are contributing factors. That is the report’s framing, not a quantified finding: it supplies no incident dataset, trend measurement or causal analysis to establish how much AI-generated code has affected outage rates.
“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, author of Building Resilient Distributed Systems
As an affiliate, we earn on qualifying purchases.
Incident Trends Remain Unquantified
The report’s claim that outages are becoming more frequent, and its suggestion that AI-assisted coding contributes to that trend, are not backed by figures, a defined time window or a comparison baseline in the supplied material. It is not possible from the report to determine whether AI-generated code has increased incident rates or how large any effect might be.
The excerpts also do not provide performance comparisons among the recovery patterns or detailed implementation guidance for particular systems. The right targets, costs and provider arrangements will vary by service; the material does not specify a single recommended configuration or establish that multi-vendor or multi-model setups will prevent failures.
As an affiliate, we earn on qualifying purchases.
Teams Must Set Their Own Targets
The next step for teams applying the framework is to identify the consequences of downtime and data loss, set RPO and RTO targets, and choose a recovery approach that fits those limits and available resources. They will also need to account for infrastructure capacity constraints and determine how to observe and respond to changes in AI models or provider availability.
The source report does not announce a follow-up study, industry standard or implementation schedule. Newman’s book, published by O’Reilly Media, is the next source for readers seeking the full treatment; the published excerpts cover only selected material from its chapters.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the news development?
The Pragmatic Engineer published authorized excerpts from Sam Newman’s 2026 book, Building Resilient Distributed Systems, with material on disaster recovery and AI software resilience.
What do RPO and RTO mean?
RPO is the maximum acceptable period of potential data loss, based on the last backup or synchronization. RTO is the acceptable time for restoring service after an incident.
Does the report prove AI-generated code is causing more outages?
No. The report’s introduction suggests AI-assisted code deployments and production bugs are contributing to outages, but the supplied material includes no data or causal analysis to verify that claim.
What AI-related risks does the excerpt address?
It discusses dependence on model providers, changes to models and the need for observability. The report also says the book considers multiple models or vendors, transparent costs and regular vendor reviews.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
