A Glimpse of the Matrix (Extended Version): Scalability Issues of a New Message-Oriented Data Synchronization Middleware

TL;DR

This paper examines the scalability challenges of Matrix, a message-oriented middleware for decentralized apps, by analyzing its federation structure and network load distribution, revealing key bottlenecks and server distribution patterns.

Contribution

It provides the first detailed analysis of Matrix's federation structure and network load, identifying specific scalability issues in its group communication mechanism.

Findings

Users are concentrated on a few large servers.

More small servers are present than previously expected.

Scalability issues are linked to the federation's structural diversity.

Abstract

Matrix is a new message-oriented data synchronization middleware, used as a federated platform for near real-time decentralized applications. It features a novel approach for inter-server communication based on synchronizing message history by using a replicated data structure. We measured the structure of public parts in the Matrix federation as a basis to analyze the middleware's scalability. We confirm that users are currently cumulated on a single large server, but find more small servers than expected. We then analyze network load distribution in the measured structure and identify scalability issues of Matrix' group communication mechanism in structurally diverse federations.