Comment on Lemmy.zip Server Update September 2025

• MrKaplan@lemmy.world ⁨2⁩ ⁨weeks⁩ ago

  more or less.

  you have 3 backend server components. scheduled tasks only run once, but there is not much performance impact from these. api you can run as many as you want, which is especially useful for rolling restarts. federation is currently the most cpu consuming part on our end. it can be split into multiple containers, but this will increase overhead and result in more total cpu usage.

  then you have a postgres database, which needs to be managed independently. i haven’t looked into viability of postgres sharding, but you’d generally just want to run the full database on a single system, possibly with replication. i don’t think lemmy can make use of read replicas currently.

  the last part is the default ui, which you can also run as many instances as you want for.

  source