Software Engineering Radio - The Podcast for Professional Software Developers · SE-Radio Team

Qian Li on DBOS Durable Execution/Serverless Computing Platform

·52 min·5 clips

Chan Lee reveals how DBOS uses idempotency keys to ensure Stripe payments are charged only once, preventing costly duplicates.

The conversation keeps its hands on the machinery. After the usual SE Radio setup, the host introduces the guest as a D-Boss co-founder and chief architect whose research dealt with efficient, reliable cloud computing abstractions. The definitions come early. Durable execution means keeping workflows reliable across systems, including when messages arrive from infrastructure such as Kafka. The stakes are plain enough. If customer data is wrong, the guest says, customers can be lost. D-Boss makes a strong claim there: when a Kafka message is received, the platform is described as making sure the message shows up across all systems. The host pushes on that. Why not write the logic directly, or use code generators to wrap durable execution around existing services? The answer moves toward guarantees and visibility. D-Boss is less about saving syntax and more about giving teams something they can operate. Postgres is the anchor. Since workflow information is stored in Postgres, customers can query what ran, what failed, and what happened step by step. That changes debugging. Execution state is no longer some black box inside a runtime. Teams can inspect structured records with SQL and reuse the same data for visualization. The guest then points to the next layer. D-Boss is building graph views for workflow steps and parent-child relationships when one workflow starts another. Management comes from the same storage model. A workflow can be resumed by re-enqueuing it, canceled by marking it canceled, or restarted against newer code with copied input and a new idempotency key. The interview stays measured and practical, then closes with dboss.dev and the guest's personal site and social profiles for follow-up.

As heard by us

A grounded look at durable workflows, Kafka delivery guarantees, and Postgres-based operational visibility.

The episode makes durable execution feel practical by asking whether a workflow can hold together across the systems that rely on it. Chan Lee, D-Boss's co-founder and chief architect, connects the platform to a Stanford-MIT research collaboration that began in 2020, then brings…

Read the full review in PlayNext →

Why you'd press play

Want to see how durable workflows stay visible and controllable in Postgres? Start here.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Software Engineering Radio - The Podcast for Professional Software Developers