扩展到数百万用户需要 Synapse Pro

Synapse Pro vs Community Synapse: Since launching Synapse Pro for nation-scale Matrix deployments, many questions arose. Community Synapse is for the Matrix community to run small to large homeservers but not for commercial hosting providers serving huge deployments. It has scalability issues with Python worker processes using at most 1 core and spending 25% of CPU on maintaining caches. In contrast, Synapse Pro:
- Reimplements worker processes in Rust to use multiple CPU cores and benefit from Rust's performance, with an idle worker using less than 3% CPU and no need for multiple workers on a 64-core server, providing >500x scalability improvement.
- Supports elastic scaling, allowing workers to be added or removed without server disruption and enabling high availability.
- Shares data between Rust workers to minimize RAM footprint and server costs.
- Enables other upcoming optimizations for even better scalability.
Memory Impact Example: The memory impact of switching Federation Reader workers from Synapse to Synapse Pro at Matrix.org last year to reduce hardware costs was shown.
Focus and Future: Community Synapse is Element's primary focus with new features first, but it's not designed for nation-scale deployments. Synapse Pro delivers a different architecture with features like worker scalability. Governments should invest in supported distributions like Element Server Suite using Synapse Pro for huge deployments.
Edit (Jan 20th 2025): To clarify, core work in Synapse will remain open source. Element will improve algorithmic complexity, database schema efficiency, etc. in community Synapse, and development will be driven by funding from Synapse Pro. Open source Synapse is the primary project, and Synapse Pro features are for removing scaling bottlenecks and high-density multi-tenancy.