The simplest way to build resilient applications
summary
Demonstrates using Restate’s Durable Execution Engine to make Java + Spring code resilient by turning steps into a persistent recovery log: live demos show sagas with compensations, awakable durable future timers for email verification, Kafka-driven long-running claims processed in parallel, and stateful “virtual objects” for package tracking. Restate runs as a single Rust binary, snapshots to S3, offers six SDKs, and reports ~15 ms median (P99 <100 ms) and ~17k throughput on a three-node, three-way-replicated cluster; repo restate-conference-demo-2025 contains the examples.
AI-written summary · about
chapters (11)
- 0:00Resilient applications intro
- 2:32Traditional resiliency challenges
- 7:12Durable execution with Restate
- 11:53How durable execution works
- 15:50Sagas and rollback
- 20:06Durable futures for async
- 24:43Long workflows with Kafka
- 28:58Stateful entities and machines
- 33:18Simpler resiliency with durable execution
- 36:05Restate server architecture
- 40:02Q&A
related talks
45mThe simplest way to build resilient applications
44mThe Simplest Way to Build Resilient Applications
27mThe simplest way to build resilient applications
35mBuilding resilient applications with Durable Execution
28mStateful Applications in the Age of Durable Execution
01h 01mLost in transaction Strategies to manage consistency in distributed systems
48mLost in transaction?
29mLet's replicate state together