Architecture & CloudLightning talk20min
Antifragile Backends: Designing Distributed Systems That Benefit From Chaos
This talk introduces antifragile backend design: systems that don’t just withstand failure, but use stress to trigger self-protection. It covers practical patterns—circuit breakers, load shedding, and backpressure—to prevent localized failures from cascading and to keep critical API and distributed system operations running.
talk.summaryAiDisclaimer
Houcine Gahboub
Most distributed systems are built to be robust, they stubbornly resist failure until they eventually break under the weight of traffic or failing microservices. But what if a system could actually benefit from chaos?
Inspired by Nassim Taleb's concept of "Antifragility," this 15-minute quickie explores how to design backend architectures that don't just survive stress, but use it to trigger self-preservation mechanisms. Instead of building rigid systems that hope things won't fail, we will look at how to implement dynamic, practical patterns that accept failure as a given. I will break down how to stop localized issues from causing cascading system crashes using straightforward architectural decisions.
Key Takeaways
This talk is for anyone building APIs or distributed systems who wants practical, low-level ways to handle failure.
Inspired by Nassim Taleb's concept of "Antifragility," this 15-minute quickie explores how to design backend architectures that don't just survive stress, but use it to trigger self-preservation mechanisms. Instead of building rigid systems that hope things won't fail, we will look at how to implement dynamic, practical patterns that accept failure as a given. I will break down how to stop localized issues from causing cascading system crashes using straightforward architectural decisions.
Key Takeaways
- The Antifragile Mindset: Understanding the practical difference between fragile, robust, and antifragile software systems.
- Circuit Breakers: How to build "electrical panels" for your microservices that trip and return errors instantly, rather than letting requests pile up and crash the whole system.
- Load Shedding: How to intentionally drop low-priority background requests to guarantee critical tasks (like checkouts) survive massive CPU spikes.
- 5 minute demo
This talk is for anyone building APIs or distributed systems who wants practical, low-level ways to handle failure.