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 options, infrastructure constraints and the risks of relying on changing models and vendors.
The Pragmatic Engineer has published excerpts from Sam Newman’s 2026 book, Building Resilient Distributed Systems, outlining how software teams can plan for large-scale outages and risks tied to AI-powered services. The material covers disaster recovery targets and patterns, as well as model-provider reliability, changing models and the need for observability.
In the excerpt from Chapter 10, Newman describes disaster recovery (DR) as the procedures used when a major event disrupts a large part of a system. Examples include fire, flooding, power-supply problems and failures in underlying cloud infrastructure. Such events can overwhelm protections aimed at keeping individual service instances running.
Newman recommends defining two measurable recovery targets. A recovery point objective (RPO) sets the maximum acceptable period of potential data loss, based on when information was last backed up or synchronized. A recovery time objective (RTO) sets how long a service can remain unavailable. Restoring data, obtaining computing resources and redirecting traffic all affect the time needed to recover.
The report also lists several DR approaches: backup and restore, pilot-light failover, warm failover and multisite active-active operation. They differ in their costs, complexity and expected recovery performance. The excerpts also address infrastructure availability, noting that teams planning recovery should account for constraints such as GPU and CPU shortages and prioritize critical services.
Recovery Targets Shape Outage Plans
The guidance matters because a recovery plan is not just a promise to bring services back: it requires decisions about acceptable downtime, possible data loss, cost and operational complexity. RPO and RTO give teams concrete measures for making those choices and testing whether a recovery approach fits the needs of a particular system.
AI services add dependencies that can complicate those plans. The report says the book addresses model-provider resilience, model changes and observability, alongside tactics for handling uncertainty such as using multiple models or vendors, making costs transparent and regularly reviewing vendors. These are presented as planning considerations, not evidence that any particular configuration guarantees resilience.
For organizations running AI-powered software, outages or provider changes can affect service behavior as well as availability. Planning for alternatives and monitoring system behavior can help teams understand where their dependencies lie. The source does not provide incident statistics or quantify how much these practices reduce failures.
As an affiliate, we earn on qualifying purchases.
From Site Failures to AI Dependencies
The published material is an excerpt from Sam Newman’s book, rather than a report of a specific outage or a new industry measurement. The Pragmatic Engineer says Newman’s book took three years to complete rather than the two he had originally planned, and that he reworked sections to address AI’s relevance to resilient system design.
Newman uses “site” as a general term for the location of computing infrastructure, whether that means a public-cloud region, a private cloud or an on-premises data center. The excerpt says the same broad disaster-recovery patterns can apply across those environments, while noting that using multiple cloud vendors raises additional issues discussed elsewhere in the book.
The newsletter says its excerpts cover selected topics across two chapters, not the full book. The supplied passage includes the start of Chapter 10’s disaster-recovery discussion and describes the broader subjects covered, including resilience for AI-powered software and ways to manage vendor uncertainty.
“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 an excerpt from Building Resilient Distributed Systems
As an affiliate, we earn on qualifying purchases.
Recovery Needs Depend on Each System
The excerpts do not prescribe a single DR approach or specify universal RPO and RTO values. Newman’s discussion presents the choice as dependent on a system’s requirements and the trade-offs among cost, complexity and recovery. The supplied text does not include detailed implementation guidance for each failover pattern.
It also does not quantify the frequency of outages, establish that AI-generated code has caused a general increase in incidents, or provide measured outcomes for the AI-resilience practices discussed. The newsletter’s introductory framing links more code deployments and bugs with concerns about production incidents, but the material supplied here offers no data to test that relationship.
The excerpt identifies model changes, provider reliability and infrastructure shortages as planning concerns, but does not name specific providers, models or shortages, or state how often these risks occur. How an organization should respond will depend on its system and vendor arrangements.
As an affiliate, we earn on qualifying purchases.
Teams Must Set and Test Targets
The practical next step described by the excerpt is for teams to define acceptable data loss and downtime, then choose recovery arrangements that match those targets. That means accounting for the full restoration process, including data recovery, access to compute capacity and traffic redirection, rather than considering only whether a backup exists.
For AI-powered services, the report points to ongoing work around observability, model and vendor options, cost transparency and vendor review. It does not announce a follow-up report, a scheduled event or a specific test program. The book, published by O’Reilly Media, contains 16 chapters; the newsletter says the excerpts cover only a portion of its material.
distributed system resilience tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the news?
The Pragmatic Engineer published excerpts from Sam Newman’s 2026 book, Building Resilient Distributed Systems, focused on disaster recovery and resilience for AI-powered software.
What do RPO and RTO measure?
RPO measures the maximum acceptable period of potential data loss, while RTO measures how long a service can be unavailable after an incident.
Which disaster-recovery approaches are mentioned?
The excerpt lists backup and restore, pilot-light failover, warm failover and multisite active-active operation. It describes these as options with different cost, complexity and recovery characteristics.
What AI-related risks does the report address?
It identifies model-provider resilience, models changing over time, observability and vendor uncertainty as issues for teams building AI-powered software. It also describes multiple models or vendors, cost transparency and regular vendor analysis as possible planning tactics.
Does the excerpt recommend one recovery plan for every system?
No. Newman presents recovery planning as a trade-off among acceptable downtime, data loss, cost and complexity. The supplied material does not give universal RPO or RTO targets.
Source: rss
Columbus Day / Indigenous Peoples' Day Picks
long weekend sales
As an affiliate, we earn on qualifying purchases.
