A Series of Unstoppable Events: Building Event-Driven Systems in Rust

Pitch

In this talk, we build event-driven systems from first principles in Rust. We’ll understand some of the core mechanics that make events unstoppable in production: flow control and backpressure under load, idempotency under repeated delivery, and predictable recovery

Description

Event-driven systems are everywhere, but “event-driven” means different things depending on who you ask. This talk cuts through the ambiguity to build a practical mental model that connects messaging, streaming, event sourcing, and asynchronous workflows, and helps us reason about these systems at scale.

From there, we go deeper into:

• Protocol Buffers for stable, language-agnostic event schemas
• State machines in Rust: ownership boundaries and concurrency
• Flow control in practice: bounded queues, backpressure, and lag
• Correctness: ordering, retries, duplicates, and idempotency
• How these primitives show up in real event platforms

We’ll look at these ideas through practical examples in Rust, including failure modes, to see how the system behaves under load, repeated delivery, and restarts.

This session is for engineers familiar with events who want a deeper understanding of building and maintaining event-driven systems.

Takeaways

• A coherent view of event-driven architecture: how events, handlers, and derived state fit together across messaging, streaming, event sourcing, and asynchronous workflows.

• A clear understanding of why Rust fits event-driven systems

Bio

Shriram Balaji is a Senior Software Engineer at Microsoft, building distributed systems in the data & compute platform of Microsoft 365.