Podcast Archives - Software Engineering Daily · Podcast Archives - Software Engineering Daily

Airbnb’s Open-Source GraphQL Framework with Adam Miskiewicz

·56 min·3 clips

Airbnb has queries with 100,000 to 300,000 fields returning megabytes of data.

The episode starts with a practical problem many software teams run into: too many separate services. That setup can lead to scattered data access, duplicated business logic, and a rough developer experience. Viaduct is presented as Airbnb's answer to that problem. It is an open-source, data-oriented service mesh and GraphQL platform built around a single, highly connected central schema. The conversation stays focused on how that architecture helps at scale. Adam explains that declaring data dependencies inside the graph has been part of Viaduct from almost day one. The API has evolved over time, but the core model has stayed the same. He uses a simple field example to make the pattern easy to follow. If a full name depends on first name and last name, the graph can declare those dependencies and fetch or compute what the resolver needs. The result is that the resolver gets the right data without needing to reach across services directly. That design choice matters more as the graph grows. Adam says Airbnb's graph is huge, with roughly 25,000 types and hundreds of thousands of fields. He also says a large share of Airbnb's online data is exposed there. At that scale, the graph becomes a place where teams can often find what they need without going back to a service boundary. The effect is strongest for higher-level, presentational features. Those features can often be built entirely by calling back into the graph. The episode frames that as re-entrancy. Instead of a resolver turning into a pile of custom service calls, the graph can be used again as an internal dependency engine. That lets the platform centralize schema and execution while keeping the system practical for product teams. The conversation also makes clear that Viaduct is not just a naming exercise or a thin API layer. It is part of how Airbnb scaled its engineering organization. The discussion is technical, but it stays grounded in engineering consequences rather than abstract platform slogans. The examples keep coming back to concrete developer work. What do you call, where do you fetch it, and how much logic should live in the graph versus somewhere else? Those are the questions the episode keeps circling. It closes by inviting people to reach out if they find Viaduct interesting, even if they cannot use it directly.

As heard by us

A practical look at Airbnb's unified data graph and why it matters for GraphQL at scale.

Viaduct centers on Airbnb's open-source, data-oriented service mesh and GraphQL platform, built around a single, highly connected central schema.

Read the full review in PlayNext →

Why you'd press play

You want a real Airbnb Viaduct scaling story, not a generic platform talk.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Podcast Archives - Software Engineering Daily