Why NanoMQ
Ultra-lightweight
Booting footprint is less than 200Kb in minimum feature set.
Asynchronous I/O
Parallelize computation for IoT & MQTT with built-in actor-model.
Multi-threading
Scale-out easily to engage multiple cores with less CPU usage in the modern SMP system.
Blazing Fast
Achieve million-level TPS. Up to 10 times faster than Mosquitto on a multi-core CPU.
Cross-platform
Highly compatible and portable with any POSIX based platform.
Interoperability
Transparent and customizable data flow. EdgeOps friendly.
Quick Start
Run with docker
docker run -d --name nanomq emqx/nanomq:latest
Or build from source
git clone https://github.com/nanomq/nanomq.git cd nanomq ; git submodule update --init --recursive mkdir build && cd build ; cmake .. ; make
NanoMQ MQTT client usage
# Publish nanomq_cli pub --url <url> -t <topic> -m <message> [--help] # Subscribe nanomq_cli sub --url <url> -t <topic> [--help] # Connect nanomq_cli conn --url <url> [--help]
Key Features
Fully MQTT 5.0 Compliant
Fully support MQTT 5.0/3.1.1, Compatible with all standard open-source MQTT SDKs.
MQTT Bridges
Bridging message from edge to multiple clouds. In-line connections to Global cloud services.
Message Persistence
Preserve your business-critical data with built-in data persistence, and resume uploading automatically when the connection is restored.
Rule Engine
Unlock the flexibility of data with a SQL-based rule engine. Integrates with eKuiper to bring stream processing to the edge.
Integrability
Reduce the cost of integrating edge computing applications with event-driven WebHooks. Easy to maintain with EdgeOps friendly HTTP APIs.
All-round
Support multiple protocols such as ZeroMQ/nanomsg/NNG/WebSocket. Enabling flexible and diverse routing topology on edge with. Secure your IoT connections with TLS/SSL.
Architecture
NanoMQ is small but mighty. It is fully featured and competes in every perspective despite the tiny size.
- Platform adaptor Layer: Detects whatever hardware and OS NanoMQ inhabited and provide compatible APIs, to avoid platform-lock-in problem.
- Task Layer: It is a built-in actor model with thread-level parallelism, which make NanoMQ highly scalable on modern SMP system.
- Transport Layer: Responsible for managing TCP/UDP stream of each pipe/client, and implement Zero-copy to reduce memory usage.
- Protocol layer: Parse byte stream to MQTT message, generate MQTT events and manage in-flight window.
- Application layer: Interact with rule-engine and global topic trie. Expose all MQTT messages and events to users with a lock-free state machine.
Built by developers, for developers
Stay connected with the community and us! Ask questions, find answers, explore ideas, and share expertise.
