Managing Asynchronous APIs at Scale

When event-driven architectures are small, teams can reason about events through word-of-mouth. They know who publishes what, who consumes it, and how messages flow through the system. Teams manage their own infrastructure or raise tickets to request changes.

At enterprise scale, that model breaks down.

Architects face systemic integration problems: hard-to-discover APIs, undocumented consumers, inconsistent or poorly enforced schemas, infrastructure drift, and a lack of support for disaster recovery beyond redeployment. Message flows become opaque, choreography fragile, and evolving an event risks breaking consumers you don’t even know about.

This talk discusses the state of the art in managing asynchronous APIs in production today.

We’ll examine how specifications like AsyncAPI, alongside alternatives such as xRegistry, can be used to describe event and messaging APIs, enabling discovery, ownership, and governance across teams. We’ll look at standardised event metadata and schema using CloudEvents, and how schema registries support validation, compatibility checks, and controlled evolution.

Rather than focusing on standards in isolation, the talk explains how these tools address our architectural concerns of discovery, governance, and provisioning.

Drawing on real-world experience at Just Eat Takeaway, we’ll show how endpoint definitions can drive code generation, schema registration, and automated messaging infrastructure—giving architects visibility, leverage, and control over complex, event-driven integrations.