RobustMQmq9
Communication Infrastructure for the AI Era
One binary, one broker, no external dependencies. Native MQTT, Kafka, NATS, and AMQP on a unified storage layer — one message, consumed by any protocol.
mq9Communication layer for AI Agents · Agent mailbox — send now, receive when online→One Message, Four Protocols
Publish via MQTT, consume natively with MQTT, Kafka, NATS, and AMQP — one data store, zero bridging, zero copying
Six Scenarios, One Binary
Not protocol silos stitched together — unified at the architectural level
AI Agent Communication
Native Agent Infrastructure
- $mq9.AI.API.* subjects for Agent register, discover, and invoke
- Million-scale Topics — each Agent gets its own isolated channel
- NATS Queue Group load balancing, zero learning overhead
IoT Device Ingestion
MQTT in / Kafka out
- Devices publish MQTT; AI platforms consume Kafka — same data
- One system replaces the MQTT + Kafka dual-broker setup
- Tiny memory footprint for edge gateway deployment
Real-Time Streaming
Full Kafka Compatibility
- Standard Kafka SDK connects directly — zero migration cost
- Million-scale lightweight Topics for large-scale partitioning
- Automatic cold data tiering to object storage
Edge-to-Cloud Sync
Unified Edge and Cloud
- Single binary, tiny footprint — deploy on edge nodes with zero dependencies
- Offline local buffering, auto-sync to cloud on reconnect
- Unified data path for factory, retail, and vehicle scenarios
Ultra-Low-Latency Dispatch
NATS Pure In-Memory Routing
- No disk writes — in-memory routing, millisecond to sub-millisecond latency
- Financial feeds, game sync, industrial control, AI inference distribution
- Switch to JetStream mode for persistence — unified storage takes over
Traditional Messaging
Native AMQP Support
- Full AMQP protocol — Exchange, Queue, Binding natively implemented
- Existing RabbitMQ applications migrate at low cost
- Gain multi-protocol interoperability at the same time
Three Components, Clear Boundaries
Compute, storage, and coordination fully separated — each layer scales independently, delivered as a single binary
Designed from the Core
Rust-Native
No GC, stable and predictable memory, no periodic spikes — consistent from edge devices to cloud clusters
Unified Storage Layer
All protocols share one storage engine — data written once, consumed by any protocol, no duplication
Native Multi-Protocol
MQTT 3.1/3.1.1/5.0, Kafka, NATS, AMQP natively implemented — full protocol semantics, not emulated
Edge-to-Cloud
Single binary, zero dependencies, offline buffering with auto-sync — same runtime from edge to cloud
AI Agent Communication
NATS-based $mq9.AI.API.* extension — native Agent registration, discovery, invocation, and orchestration
Multi-Mode Storage
Memory / RocksDB / File, per-Topic configuration, automatic cold data tiering to S3
Shared Subscription
Break the "concurrency = partition count" limit — consumers scale elastically at any time
Native Multi-Tenancy
Unified multi-tenancy across all protocols — full data isolation and independent permissions per tenant
Minimal Operations
Single binary, zero external dependencies, built-in Raft consensus, ready out of the box
Start a Cluster in 3 Steps
Single binary, zero external dependencies, one command per node