Experimental Analysis of Efficiency of the Messaging Layer Security for Multiple Delivery Services

TL;DR

This work provides an empirical assessment of MLS/CGKA efficiency by implementing a configurable MLS testbed with two Delivery Services (MQTT and GossipSub) and a simulated client population. It reveals that practical costs scale near linearly with group size in common scenarios and that the Delivery Service and update paradigm have substantial impact on performance, with Welcome and GroupInfo messages dominating bandwidth. The study introduces a formal Average Update Cost metric and analyzes the effects of External Joins and the ratchet-tree state on generation and processing times. The results challenge the assumption of logarithmic scaling in real deployments and offer a reusable open-source platform for further exploration of MLS performance under diverse conditions.

Abstract

Messaging Layer security (MLS) and its underlying Continuous Group Key Agreement (CGKA) protocol allows a group of users to share a cryptographic secret in a dynamic manner, such that the secret is modified in member insertions and deletions. One of the most relevant contributions of MLS is its efficiency, as its communication cost scales logarithmically with the number of members. However, this claim has only been analysed in theoretical models and thus it is unclear how efficient MLS is in real-world scenarios. Furthermore, practical decisions such as the chosen Delivery Service and paradigm can also influence the efficiency and evolution of an MLS group. In this work we analyse MLS from an empirical viewpoint: we provide real-world measurements for metrics such as commit generation and processing times and message sizes under different conditions. In order to obtain these results we have developed a highly configurable environment for empirical evaluations of MLS through the simulation of MLS clients. Among other findings, our results show that computation costs scale linearly in practical scenarios even in the best-case scenario.

Figures (10)

• Figure 1: Message flow in MLS through the Delivery Service.
• Figure 2: Overview of the Invitation and External Join methods through which new members are added to the MLS group.
• Figure 3: Subscription process for new group members.
• Figure 4: Example Client configuration file.
• Figure 5: Logs generated by clients.