138: Tom Preston-Werner - Building Full-Stack JS Apps with Redwood.js

April 22, 2020

73 min episode · 2 min read

Tom Preston-Werner

Episode

73 min

Read time

2 min

AI-Generated Summary

Published Dec 25, 2025

Key Takeaways

✓Router Architecture: Redwood uses a single flat routing file instead of nested routing, making it simple to trace URLs to page components. All routes live in one place for easier debugging and long-term maintenance, though this requires solving layout persistence challenges with technical solutions under the hood.
✓Cells Pattern: Cells provide declarative data fetching by exporting query, success, failure, and empty state components. This abstraction allows Redwood to transparently solve problems like the GraphQL waterfall issue where nested components wait unnecessarily for parent queries to complete before fetching their own data in parallel.
✓Services Layer: Services sit above GraphQL resolvers as the primary back-end API, organizing business logic by domain (billing, content, accounts) rather than database tables. Functions export directly from service files, enabling reuse across services without resolver boilerplate while maintaining clear boundaries and encapsulation for maintainability.
✓Serverless Database Strategy: Redwood deploys GraphQL APIs as single Lambda functions, accepting current connection pooling limitations. The roadmap targets distributed databases like Yugabyte or Fauna for edge-ready deployments, with AWS RDS Proxy and Netlify partnerships addressing near-term scaling needs without requiring Kubernetes or server management.
✓Tutorial-Driven Development: Redwood's development methodology writes comprehensive tutorials first, then builds features to make tutorials work. This ensures the framework delivers the intended developer experience from day one, with the tutorial serving as the primary introduction and validation of core functionality for new users.

What It Covers

Tom Preston-Werner introduces Redwood.js, a full-stack JavaScript framework designed as a Rails replacement that deploys serverless, uses GraphQL APIs, React front-ends, and Prisma for database access with strong opinions on maintainability.

Key Questions Answered

•Router Architecture: Redwood uses a single flat routing file instead of nested routing, making it simple to trace URLs to page components. All routes live in one place for easier debugging and long-term maintenance, though this requires solving layout persistence challenges with technical solutions under the hood.
•Cells Pattern: Cells provide declarative data fetching by exporting query, success, failure, and empty state components. This abstraction allows Redwood to transparently solve problems like the GraphQL waterfall issue where nested components wait unnecessarily for parent queries to complete before fetching their own data in parallel.
•Services Layer: Services sit above GraphQL resolvers as the primary back-end API, organizing business logic by domain (billing, content, accounts) rather than database tables. Functions export directly from service files, enabling reuse across services without resolver boilerplate while maintaining clear boundaries and encapsulation for maintainability.
•Serverless Database Strategy: Redwood deploys GraphQL APIs as single Lambda functions, accepting current connection pooling limitations. The roadmap targets distributed databases like Yugabyte or Fauna for edge-ready deployments, with AWS RDS Proxy and Netlify partnerships addressing near-term scaling needs without requiring Kubernetes or server management.
•Tutorial-Driven Development: Redwood's development methodology writes comprehensive tutorials first, then builds features to make tutorials work. This ensures the framework delivers the intended developer experience from day one, with the tutorial serving as the primary introduction and validation of core functionality for new users.

Notable Moment

Preston-Werner reveals his frustration with JavaScript tooling led him to build Redwood after finding no suitable ORM until Prisma 2 emerged. He invested in Prisma after choosing it, believing their query builder finally matched Active Record's developer experience in the JavaScript ecosystem.

Know someone who'd find this useful?