Understanding messaging protocols: XMPP and Matrix

In the world of real-time communication, two prominent protocols often come into discussion: XMPP and Matrix. Both protocols aim to provide robust and secure messaging solutions, but they differ in architecture, features, and community adoption. This article delves into the key differences and similarities between XMPP and Matrix to help you understand which might be better suited for your needs.

What is XMPP?

Overview

XMPP (Extensible Messaging and Presence Protocol) is an open-standard communication protocol originally developed for instant messaging (IM). It was designed as the Jabber protocol in 1999 to aggregate communication across a number of options, such as ICQ, Yahoo Messenger, and MSN. It was standardized by the IETF as RFC 3920 and RFC 3921 in 2004, and later revised as RFC 6120 and RFC 6121 in 2011.

Key Features

• Decentralized Architecture: XMPP operates on a decentralized network of servers. The protocol is said to be federated. The network of all interconnected XMPP servers is called the XMPP federation.
• Extensibility: The protocol is highly extensible through XMPP Extension Protocols (XEPs). There are currently more than 400 extensions covering a broad range of use cases like social networking and Internet of Things features through PubSub extensions, Groupchat (aka MUC, Multi-user chat), and VoIP with the Jingle protocol.
• Security: Supports TLS for encryption and SASL for authentication. End-to-end encryption is available through the OMEMO extension.
• Interoperability: Widely adopted with numerous clients and servers available.
• Gateways: Built-in support for gateways to other protocols, allowing for communication across different messaging systems.

Network Protocol Design

• TCP-Level Stream Protocol: XMPP is based on a TCP-level stream protocol using XML and namespaces. This extensibility while maintaining schema consistency is key. It can also run on top of other protocols like WebSocket or HTTP through the concept of binding.

Use Cases

• Instant messaging
• Presence information
• Multi-user chat (MUC)
• Social networks
• Voice and video calls (with extensions)
• Internet of Things
• Massive messaging (massive scale messaging platforms like WhatsApp)

What is Matrix?

Overview

Matrix is an open standard protocol for real-time communication, designed to provide interoperability between different messaging systems. It was introduced in 2014 by the Matrix.org Foundation.

Key Features

• Decentralized Architecture: Like XMPP, Matrix is also decentralized and supports a federated model.
• Event-Based Model: Uses an event-based architecture where all communications are stored in a distributed database. The conversations are replicated on all servers in the federation that participate in the discussion.
• End-to-End Encryption: Built-in end-to-end encryption using the Olm and Megolm libraries.
• Bridging: Strong focus on bridging to other communication systems like Slack, IRC, and XMPP.

Network Protocol Design

• HTTP-Based Protocol: Matrix uses HTTP for communication and JSON for its data structure, making it suitable for web environments and easy to integrate with web technologies.

Use Cases

• Instant messaging
• VoIP and video conferencing
• Bridging different chat systems

Detailled Comparison

Architecture

• XMPP: Uses a federated model to build a network of communication that works for both messaging and social networking. The content is not duplicated by default.
• Matrix: Uses a federated model where each server stores a complete history of conversations, allowing for decentralized control and redundancy.

XMPP is built around an event-based architecture to reach the largest possible scale. Matrix is built around a distributed model that may be more appealing to smaller community servers. As the conversations are distributed, it can cope more easily with servers suffering from frequent disconnections in the federated network.

Extensibility

• XMPP: Extensible through XEPs that are standardized by the XMPP Standards Foundation, allowing for a wide variety of additional features. As the protocol is based on XML, it can also be extended for custom client features, using your own namespace. The XML schema can be used to define your extension data structure.
• Matrix: Extensible through modules and APIs, with a strong focus on bridging to other protocols. It is extensible as well and allows custom events and custom properties.

Security

• XMPP: Supports TLS for secure communication and SASL for authentication. End-to-end encryption is available through extensions like OMEMO.
• Matrix: Supports TLS for secure communication. Built-in end-to-end encryption using Olm and Megolm, providing robust security out of the box.

Both end-to-end encryption approaches are similar, as they are both based on the same double ratchet encryption algorithm made popular by the Signal messaging platform.

Interoperability

• XMPP: Known for its interoperability due to its long-standing presence and wide adoption. Includes built-in support for gateways to other protocols.
• Matrix: Designed with interoperability in mind, with native support for bridging to other protocols. More recent gateways are available. They could be ported to work on both protocols (which would be neat).

Scalability

• XMPP: By design, XMPP has an edge in terms of scalability. XMPP is event-based and works as a broadcast hub for messages, making it efficient in handling a large number of concurrent users. It is proven to sustain millions of concurrent users.
• Matrix: Matrix maps conversations to documents that are replicated across servers involved in the discussion. This means the document state needs to be merged and reconciled for each new posted message, which incurs significant overhead in terms of processing power, memory, and storage. Its use case is mainly “organization level” chat, supporting thousands of users, not millions.

Community and Adoption

• XMPP: Established and widely adopted with a large number of client and server implementations. This can be seen as a drawback, leading to intimidating choices of tools. However, this has proven to be a strength with many competing implementations that have proven to be interoperable. This is a validation of the robustness of the protocol. Initially developed by Jeremy Miller, he cocreated Jabber, Inc to support the first server. The company was later acquired by Cisco. It is now an Internet Engineering Task Force standard used for massive scale deployments and a protocol drive by the non-profit XMPP Standard Foundation.
• Matrix: Rapidly growing community with increasing adoption, particularly in open-source projects and decentralized applications. The main implementation is developed by Element, the company funded to grow the Matrix protocol.

Conclusion

Both XMPP and Matrix offer robust solutions for real-time communication with their own strengths. XMPP’s long history, extensibility, and efficient scalability make it a reliable choice for traditional instant messaging and presence-based applications, but also social networks, Internet of Things, and workflows that mix human users and devices. On the other hand, Matrix’s architecture, built-in end-to-end encryption, and focus on gateway development make it an excellent choice for those looking to integrate multiple communication systems or require secure corporate messaging through the Element client.

Using a server like ejabberd is a future-proof approach, as it is multiprotocol by design. ejabberd supports XMPP, MQTT, SIP, can act as a VoIP and video call proxy (STUN/TURN), and can federate with the Matrix network. It is likely to support the Matrix client protocol as well in beta in the near future.

Choosing between XMPP and Matrix depends largely on your specific needs, existing infrastructure, and future scalability requirements. Both protocols continue to evolve, offering exciting possibilities for real-time communication.

Mistakes? If you spot a mistake, please reach out to share it! Thanks! I would like this document to be as accurate as possible.