@xmpp @peter @patrik

"Matrix server is not designed for such a load."

can this be changed in the future, or is this a fundamental part of the architecture of ?

Not knowing much about the technique behind it, I have the feeling that is more designed for many small servers.

This scenario does change some issues you have to count on if you think of some big servers.

Please tell me if I´m right or wrong?

@Nachbarschaft @patrik @xmpp

Matrix can run on small-ish servers although currently the RAM usage is about 1.5GB. That should be significantly reduced when the python3 version of Synapse is out, and it will probably become feasible to run on a Raspberry Pi.

The "one giant server" approach of is a problem. They need to break it up into multiple servers and be much more proactive about encouraging people to run servers. A similar problem applies to

@bob @patrik fwiw py3 is out and usable today and reduces RAM by 2-3x. Meanwhile we're working on improving migration/portability to be able to break up

I have no idea where the "4 core and 4GB is not enough of 700" FUD is coming from; with 4 cores and 4GB you should be able to support at least 5K simultaneous users, depending on the complexity of the rooms.

@matrix @patrik That's good. It means that a matrix server should be hostable on a board with 1GB RAM without too much nail biting.
Sign in to participate in the conversation
Moin & Herzlich Willkommen

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!