TL;DR
Get monitors, keyboards and dev gear delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
The Pragmatic Engineer podcast published an episode featuring software architect Sam Newman discussing his new book, Building Resilient Distributed Systems. Newman outlines common causes of distributed-system failures and discusses microservices, observability, business trade-offs and how AI may affect software teams.
The Pragmatic Engineer has published a podcast episode in which Sam Newman discusses resilient distributed systems, including lessons from his new book, Building Resilient Distributed Systems. The conversation covers why services fail, how teams should think about microservices and observability, and how software developers can use AI without losing understanding of the systems they build.
Newman identifies three recurring constraints in distributed systems: information takes time to travel, services or resources can become unavailable, and computing resources are finite. He says that, in his experience, many outages occur when resource pools run out, including capacity for CPU, memory, storage or network traffic. These are observations he shares in the episode, not a quantified industry-wide outage analysis.
The discussion also addresses idempotency, a way to prevent repeating an operation from producing an unintended second effect. Newman uses payments as an example: retrying a request should not charge a customer twice. The source describes two approaches: clients can send unique idempotency keys for servers to track, or servers can generate fingerprints from request details. It says keys are straightforward to build but harder to add to an existing system because both client and server may need changes; fingerprints can be added on the server side, but may mistakenly treat legitimate repeat requests as duplicates.
Other topics include the role of observability in understanding system behavior and the need to make error-handling decisions in light of business consequences. The episode also considers AI-assisted development, including whether specifications or code should serve as the source of truth, and the risks Newman calls “cognitive debt” and “cognitive surrender.” The source says modular architecture can give teams room to experiment with AI while retaining a grasp of the systems they develop.
Designing for Failure and Capacity
The conversation focuses on practical failure conditions that can affect services used by businesses and their customers: delays, outages and exhausted capacity. Retries and finite resources can turn a local problem into a wider one, while poorly handled repeated requests can have direct consequences such as duplicate payments. The episode frames resilience as a matter of system design and operational judgment, rather than assuming that distributed components will remain available.
Newman’s discussion of business context also matters because a system’s response to failure can affect users differently. The source says the episode considers when to fail open or fail closed, but does not prescribe one answer for every service. That choice depends on what a particular system protects and what an interruption or permissive fallback would mean for the business and its users.
For teams adopting AI tools, the episode raises a related concern: faster code production does not automatically preserve the team’s understanding of the system. Newman’s emphasis on modular architecture and observability points to ways teams can retain clearer boundaries and see how software behaves as they experiment.
distributed system monitoring tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Newman’s Microservices Perspective
Newman is the author of Building Microservices, but the source says he describes microservices as an architecture of “last resort.” In the episode, he offers a clear definition based on whether a service can be deployed independently, without requiring other services to be deployed at the same time. A softer definition describes services organized around business functions rather than technical layers.
The source also recounts the term’s history. Newman was present at an architecture symposium in England in the early 2010s where the term “microservices” was suggested during a discussion of independently replaceable services. James Lewis and Martin Fowler published an article defining the architectural term in March 2014, and Newman’s book followed in 2015. The episode revisits this history to explain why Newman argues that independent deployment and team autonomy matter more than adopting a label.
The Pragmatic Engineer page says the episode is available on YouTube, Apple and Spotify, and includes a transcript and timestamps. It also promotes a separate webinar on property-based testing and announcements from technology companies; those items are not presented as findings from Newman’s discussion.
““last resort””
— Sam Newman, as quoted in The Pragmatic Engineer’s episode summary
microservices observability software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Details Beyond the Episode Summary
The supplied source is an episode summary rather than the full transcript, and it does not state the episode’s publication date. It gives no measured results for Newman’s discussion of outages, observability or AI-assisted development; his statement that resource exhaustion causes many outages is presented as his experience, without a dataset or defined comparison period.
The summary’s description of fingerprint-based idempotency ends mid-sentence after noting that legitimate requests can be rejected as duplicates. It does not provide the rest of that explanation or further implementation detail. The source also does not identify specific systems or organizations examined as case studies, or report a new research study, product launch or formal standard resulting from the conversation.
As an affiliate, we earn on qualifying purchases.
Where to Hear the Conversation
Readers can listen to or watch the episode through the YouTube, Apple and Spotify links listed on The Pragmatic Engineer page. The page also provides the episode transcript and timestamps for locating particular topics. No follow-up publication, new release date or further development is specified in the source material.
The next useful step for readers seeking technical detail is to consult the full episode and transcript, especially for Newman’s discussion of business-specific fail-open and fail-closed decisions, observability and the trade-offs between idempotency keys and request fingerprints. The available summary does not settle those questions for particular systems.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the news about Sam Newman?
The Pragmatic Engineer published a podcast episode featuring Newman on resilient distributed systems, drawing on his new book and covering microservices, system failures, observability and AI-assisted development.
What three distributed-systems rules does Newman describe?
He says information takes time to travel, the service or resource being contacted may be unavailable, and resources such as CPU, memory, storage and bandwidth are finite.
How does Newman define a microservice?
The episode summary gives a clear definition: a service that can be deployed independently without requiring other services to be deployed at the same time. It also gives a softer definition based on boundaries around business functions.
What is idempotency, and why does it matter?
Idempotency means repeating an operation has the same effect as performing it once. It can prevent unintended outcomes such as charging a customer twice after a payment request is retried.
Where can readers find the episode?
The Pragmatic Engineer says the episode is available on YouTube, Apple and Spotify, with a transcript and timestamps on its page.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
