Open Source Elixir Clustering Software

Swarm is a distributed process registry for Elixir/Erlang that lets you register and discover processes across a cluster as if they were on a single node. It focuses on automatic distribution and rebalancing, so when nodes join or leave, Swarm can move work and hand off state to keep the system stable. The library offers a global name registry, conflict resolution for name ownership, and hooks for reacting to membership changes. Its design embraces OTP principles, using supervised processes and fault-tolerant messaging to keep the registry resilient under failures. Developers can co-locate Swarm with their existing supervision trees to scale workers horizontally without writing custom clustering code. Typical use cases include globally unique workers, sharded consumers, and presence-like coordination where node churn is expected. By turning registry and handoff into configuration and callbacks, Swarm reduces the complexity of multi-node deployments.

libcluster provides automatic cluster formation and node discovery for Erlang/Elixir applications, removing the manual steps typically required to connect distributed nodes. It ships with multiple pluggable strategies—such as DNS polling, Kubernetes integration, EC2 metadata/tag discovery, UDP gossip, and EPMD helpers—so you can adapt clustering to your runtime environment. The library supervises a topology process that continually discovers peers and joins or prunes nodes as they become available or disappear. Its design embraces OTP principles: strategies implement a behavior, can be composed, and run under supervision for resilience. libcluster also emits events on nodeup/nodedown, allowing applications to react (for example, rebalance work or update registries) when membership changes. Because it relies on the BEAM’s native distribution, it integrates cleanly with Phoenix PubSub, Horde, Oban, or any other distributed Elixir tooling.