Eliminating Failure Classes in Distributed Systems with Temporal by Anton Ragutskii

Modern high-load backend systems and distributed architectures often look solid on paper — yet in production they can become brittle, difficult to reason about, and surprisingly expensive to operate. Retries pile up, partial failures turn into incidents, and long-running business processes become a maze of timeouts, compensations, and “what if” logic spread across multiple services. In this talk, Anton will share practical lessons from real systems in fintech, gaming, and enterprise platforms, and explain how durable execution and workflow-as-code approaches — powered by Temporal — can help model long-running processes in a way that is observable, testable, and resilient by design. You’ll see how this approach can eliminate entire classes of failures that teams often accept as “just how distributed systems work.” Anton Ragutskii is a Staff Engineer with 10+ years of experience designing and operating high-load backend systems and distributed architectures. Most of my work is focused on building reliable production services using Java, Spring Boot, and modern cloud-native tooling. Over the years, working with complex systems across fintech, gaming, and enterprise platforms, he has seen the same pattern repeat: architectures evolve fast, but reliability and operational clarity often degrade over time. That pushed him to look for better ways to model long-running business processes and handle failure in distributed systems. Language – English Subscribe to our channel, so you don't miss our videos on the most relevant topics from real professionals🔥Learn more about "Nine Brains" high-quality courses by world-class trainers ⬇️ https://ninebrains.eu You can follow us on: ➡️LinkedIn   / devclub-lv   ➡️Facebook   / devclublv