Discover and explore top open-source AI tools and projects—updated daily.
sockudoHigh-performance realtime platform with AI Transport
Top 43.9% on SourcePulse
Sockudo is a high-performance, self-hosted realtime platform designed as a drop-in replacement for Pusher, written in Rust. It targets developers and teams seeking a fast, scalable, and feature-rich alternative for WebSocket and HTTP publish workloads, offering both strict Pusher Protocol V1 compatibility and advanced Protocol V2 features for enhanced delivery semantics and control. Sockudo provides significant benefits through its speed, flexibility, and extensive feature set, enabling robust realtime communication infrastructure.
How It Works
Sockudo operates as a high-performance Rust server supporting WebSocket and HTTP publish workloads. It maintains strict Pusher Protocol V1 compatibility while introducing Protocol V2, which adds features like message IDs, recovery, rewind, tags, deltas, and mutable message events. The platform supports horizontal fanout through integrations with various message brokers including Redis, NATS, RabbitMQ, Google Pub/Sub, Kafka, and Pulsar. Its architecture emphasizes durable history, hot replay buffers, two-tier recovery, and versioned mutable messages, alongside presence tracking and message annotations. An optional AI Transport layer is built upon these core primitives.
Quick Start & Requirements
git clone https://github.com/sockudo/sockudo.git && cd sockudo && make uprustup toolchain install stable && git clone https://github.com/sockudo/sockudo.git && cd sockudo && cargo run -p sockudostable)../charts/sockudoHighlighted Details
until_attach history reads for robust message persistence and retrieval.Maintenance & Community
Sockudo is managed via GitHub Issues and Discussions. Community engagement is also fostered through Discord and X/Twitter. The repository structure follows a monorepo layout, housing server crates, SDKs, dashboard, and documentation.
Licensing & Compatibility
Sockudo is licensed under the MIT License. This permissive license generally allows for commercial use and integration into closed-source projects without significant restrictions, though individual SDK packages may carry their own notices.
Limitations & Caveats
The AI Transport feature is opt-in, disabled by default, and requires Protocol V2. Client and server SDKs are currently intended to be installed from local monorepo paths until package-manager publishing workflows are fully enabled. For high-traffic deployments using horizontal adapters, relying solely on memory-based history, version stores, or process-local cache may not be advisable.
1 day ago
Inactive
mistralai
openai