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 in AI-powered software. The material explains recovery targets and approaches, while emphasizing that model providers, changing models and constrained computing capacity create dependencies teams need to plan for.
The Pragmatic Engineer has published excerpts from Sam Newman’s 2026 book, Building Resilient Distributed Systems, outlining ways software teams can prepare for major outages and failures in AI-powered services. The report covers disaster recovery targets and patterns, as well as risks tied to AI model providers, shifting models and limited computing infrastructure.
Newman’s excerpt explains that disaster recovery planning begins with setting two measurable targets. The recovery point objective (RPO) is the maximum acceptable period of potential data loss, based on how recently information was backed up or synchronized. The recovery time objective (RTO) is the acceptable time a service can remain unavailable after an incident.
The excerpt discusses several recovery approaches: backup and restore, pilot-light failover, warm failover and multisite active-active systems. These options involve different costs, complexity and recovery capabilities. Newman presents the choice as dependent on a system’s requirements rather than a single approach suitable for every organization.
The report also addresses AI-specific dependencies. It highlights the resilience of model providers, the challenge of models changing and the need for observability. The suggested planning tactics include using multiple models or vendors, making costs transparent and regularly analyzing vendors. The report notes that ongoing GPU and CPU shortages can constrain infrastructure availability, making it important to understand supplier limits and prioritize critical services.
Recovery Targets Shape Outage Plans
The material matters to teams whose services depend on complex infrastructure or external AI providers because a local fix may not be enough when a whole site, region or provider becomes unavailable. Explicit RPO and RTO targets give teams a way to define acceptable data loss and downtime before a failure occurs, and to judge whether a recovery plan meets those expectations.
AI systems add dependencies that can change over time. If a model, vendor or supply of computing capacity becomes unavailable or behaves differently, an application may be affected even when its own software has not changed. Planning for provider and model uncertainty can help teams identify critical services and consider alternatives. The excerpt offers planning principles, not evidence that any particular architecture prevents outages.
As an affiliate, we earn on qualifying purchases.
From Site Failures to AI Dependencies
The report is based on excerpts from two chapters of Newman’s book, published by O’Reilly Media. Its disaster recovery chapter considers events such as fires, floods, electrical failures and problems with underlying cloud infrastructure that can affect large parts of a system at once. Newman uses “site” to refer broadly to where computing infrastructure operates, whether that is a cloud region or a data center.
The Pragmatic Engineer says Newman spent three years writing the book, rather than the two he originally planned, and revised it to account for AI. The publication frames the excerpts as practical material for engineers building reliable systems. Its report also says the recommendation was unpaid and that its links were not affiliate links.
“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.
No Prescribed Setup for Every Team
The excerpts do not specify which disaster recovery pattern a particular organization should choose or set universal RPO and RTO targets. Those decisions depend on a system’s needs, acceptable downtime and data loss, and available budget and infrastructure. The report also does not provide outage statistics or evidence comparing the effectiveness of the approaches.
Although the source says AI agents are contributing to more code deployments and bugs, it provides no supporting figures or separate attribution for that claim. The extent to which AI-related changes are driving production incidents is therefore not established by the supplied material. It also does not identify specific vendors, model changes or compute shortages affecting an individual service.
distributed system resilience tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Teams Must Set Their Own Targets
The report presents the book excerpts as guidance for organizations to apply to their own systems. The next practical step described by the material is to set acceptable data-loss and downtime targets, then evaluate recovery options against cost, complexity and available infrastructure. Teams planning AI-powered software can also review provider dependencies, observability, model changes and alternatives across vendors.
The excerpts do not announce a new standard, product or scheduled follow-up. Whether organizations adopt these practices, and how well particular recovery plans perform in real incidents, remains to be seen.
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, on disaster recovery and resilience for AI-powered software.
What do RPO and RTO mean?
RPO sets the maximum acceptable period of potential data loss. RTO sets how long a service can be unavailable after an incident.
Which disaster recovery approaches does the excerpt cover?
It discusses backup and restore, pilot-light failover, warm failover and multisite active-active. The excerpt says these approaches differ in cost, complexity and recovery capability.
What AI-related resilience concerns does the report raise?
It points to model-provider resilience, models changing, observability and constrained GPU or CPU availability. Suggested planning tactics include using multiple models or vendors and regularly reviewing vendors.
Does the report show that AI agents are causing more outages?
No. The source says increased AI-assisted code deployments and bugs are contributing to more incidents, but provides no figures or independent evidence to quantify that claim.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
