| name | zeromq-zguide-3-2-0 |
| description | Complete ZeroMQ ZGuide 3.2 toolkit covering messaging patterns, socket types, reliability mechanisms, and distributed computing architectures. Use when building asynchronous message-driven applications, implementing request-reply patterns, pub-sub systems, load balancing, or designing scalable distributed systems with ZeroMQ/3.2+. |
ZeroMQ ZGuide 3.2
Overview
ZeroMQ (also known as ØMQ, 0MQ, or zmq) is an embeddable networking library that acts like a concurrency framework. It gives you sockets that carry atomic messages across various transports like in-process, inter-process, TCP, and multicast. You can connect sockets N-to-N with patterns like fan-out, pub-sub, task distribution, and request-reply. It's fast enough to be the fabric for clustered products. Its asynchronous I/O model gives you scalable multicore applications built as asynchronous message-processing tasks.
ZeroMQ is from iMatix and is LGPLv3 open source. The ZGuide ("The Guide") is written by Pieter Hintjens and 100+ contributors, with 60+ diagrams and 750 examples in 28 languages. This version covers ZeroMQ 3.2.
Key projects in the ZeroMQ ecosystem:
- libzmq — core engine in C++, implements ZMTP/3.1 (10,800+ stars)
- czmq — high-level C binding for ØMQ
- pyzmq — Python bindings (4,100+ stars)
- cppzmq — header-only C++ binding
- netmq — 100% native C# implementation
- jeromq — pure Java implementation
- zeromq.js — Node.js bindings
- zmq.rs — native Rust implementation
- zyre — framework for proximity-based peer-to-peer applications
- majordomo — service-oriented reliable queuing protocol
- malamute — ZeroMQ enterprise messaging broker
When to Use
- Building asynchronous message-driven applications
- Implementing request-reply patterns (REQ/REP, DEALER/ROUTER)
- Creating pub-sub data distribution systems
- Designing load-balanced task distribution architectures
- Building scalable distributed systems with ZeroMQ 3.2+
- Working with multipart messages and message envelopes
- Implementing reliable messaging patterns (Pirate patterns, Majordomo)
- Designing peer-to-peer or proximity-based networking
- Creating custom protocols on top of ZeroMQ ("unprotocols")
Core Concepts
Sockets, not connections. ZeroMQ presents a familiar socket API but hides message-processing engines underneath. Sockets have types that define routing semantics — you plug them together like construction toys to define your network architecture.
Asynchronous I/O. All I/O happens in background threads. Messages arrive in local input queues and are sent from local output queues. Your application code never blocks on network I/O (except in specific cases like REQ sockets).
Messaging patterns. ZeroMQ routes and queues messages according to precise recipes called patterns. The four core patterns are:
- Request-reply — clients to services (RPC, task distribution)
- Pub-sub — publishers to subscribers (data distribution)
- Pipeline — fan-out/fan-in (parallel task distribution and collection)
- Exclusive pair — two sockets exclusively (thread-to-thread)
No broker. ZeroMQ is "zero broker" — it provides the messaging fabric without requiring a central message broker. You can build brokers from ZeroMQ itself if needed.
Transports. ZeroMQ abstracts over multiple transports: tcp (disconnected TCP, elastic and portable), ipc (inter-process, UNIX only), inproc (in-process, fastest), pgm and epgm (multicast).
Advanced Topics
Basics: Hello World, task distribution, context lifecycle, version reporting → Basics
Sockets and Patterns: Socket API, messaging patterns, multipart messages, proxies, polling, multithreading → Sockets and Patterns
Advanced Request-Reply: Reply envelopes, ROUTER identities, load balancing, async client/server → Advanced Request-Reply
Reliable Messaging: Pirate patterns, heartbeating, Majordomo protocol, Binary Star failover → Reliable Messaging
Advanced Pub-Sub: Espresso tracing, last-value caching, slow subscriber detection, Clone pattern → Advanced Pub-Sub
Architecture and Protocols: MOPED design process, unprotocols, serialization, FileMQ, SASL auth → Architecture and Protocols
Distributed Computing: Discovery, presence, peer-to-peer Harmony pattern, Zyre framework → Distributed Computing
