WolfScale is a lightweight, high-availability replication layer for MariaDB databases and clusters, written in Rust for maximum performance and reliability.

It provides automatic leader election with deterministic fail-over, ensuring predictable behaviour during outages—the node with the lowest ID always becomes leader, eliminating split-brain scenarios. WolfScale uses a Write-Ahead Log (WAL) to replicate changes across nodes, with built-in catch-up mechanisms that allow nodes to safely rejoin the cluster after failures without risking data inconsistency.

Unlike Galera Cluster, WolfScale uses a single-leader model that eliminates write conflicts and certification delays, making it simpler to operate and debug.

The included MySQL-compatible proxy transparently routes write requests to the current leader while distributing reads across followers, requiring zero application changes.

With its HTTP API, command-line tools WOLFCTL and comprehensive health monitoring, WolfScale offers a modern, operationally simple approach to MariaDB high availability that deploys as a single binary alongside your existing database infrastructure.