valkey-swift: Lessons from building a production-grade Valkey client library in Swift

Swift is a compelling choice for production backend service, with a rich open-source ecosystem that gives teams a single language across the entire stack. Every serious backend needs a fast caching layer — and until now, the Swift server ecosystem had no production-grade Valkey client.

The existing Swift Redis clients predate Swift’s modern concurrency model and lack critical Valkey features like cluster-mode support. So we built valkey-swift - a native Valkey Swift client for modern Swift services!

This talk covers the key features of valkey-swift and the design decisions behind them. It ships with full cluster-mode support, covers all available Valkey commands including pipelines, transactions, Pub/Sub and extends beyond core Valkey commands to support Valkey modules. For performance and reliability it supports load balancing, circuit breaking and connection pooling, and secures connections over TLS for enhanced security.

Beyond the feature set, this talk explores what Swift’s language model makes possible for client library design — where structured concurrency turns resource lifetimes, cancellation, and connection safety from conventions you hope users follow into guarantees the compiler enforces.

While some of these design decisions are Swift-specific, a multitude of others are generic and applicable to any language: how do you model failure modes before they surface in production, how do you keep 400+ commands maintainable over time, and how do you treat API stability as a first-class design constraint rather than an afterthought.

If you’re building backend services in Swift, this is the Valkey client library you’ve been waiting for — and if you’re not, the lessons learned building it apply to anyone designing production-grade client libraries in any language.