this post was submitted on 12 Jun 2023
402 points (99.5% liked)
Lemmy.World Announcements
29042 readers
2 users here now
This Community is intended for posts about the Lemmy.world server by the admins.
Follow us for server news ๐
Outages ๐ฅ
https://status.lemmy.world
For support with issues at Lemmy.world, go to the Lemmy.world Support community.
Support e-mail
Any support requests are best sent to info@lemmy.world e-mail.
Report contact
- DM https://lemmy.world/u/lwreport
- Email report@lemmy.world (PGP Supported)
Donations ๐
If you would like to make a donation to support the cost of running this platform, please do so at the following donation URLs.
If you can, please use / switch to Ko-Fi, it has the lowest fees for us
Join the team
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Likely experience and knowledge improving the quality of deployment. Most instances are likely underspecced, are on hosts that aren't easy to scale up with, or are maxed out in their current offering tier (lemmy.ml comes to mind there)
I wouldn't be surprised if it has more to do with caching than throwing hardware at it.
Looking at @ruud@lemmy.world's post, he moved the instance to a pretty beefy server - it sounds like a large part of the stability is coming from overestimating performance requirements.
* correctly estimating
๐
Correct. Lemmy is a monolithic application, so there's only so much a server upgrade can do.
This is sort of true, but not really true. The default docker setup is comprised of 4 containers. I've seen admins report that two of those containers (
lemmy
andlemmy-ui
) can be horizontally scaled just fine. Thepict-rs
andpostgres
containers can currently only be vertically scaled, but Postgres natively supports scaling read load.at least through read-replicas, and there's an incomplete proposal to support scaling reads through separate db connections.All of which is to say, it's possible to throw 4-6 machines at a Lemmy install. It's not truly a single-procees monolith. Would the Lemmy code be able to productively use all that hardware? I dunno. It's scaled better tombig hardware on
lemmy.world
than I would have predicted last week, maybe it can fully utilize a 6 machine setup, or maybe the db falls over first and you need to fix performance bugs because sn instance can scale to the user counts necessary to support bigger hardware setups.I wonder if you could replace Postgres with CockroachDB in this instance to scale out.
What is Pict-rs?
The image-hosting component: https://crates.io/crates/pict-rs
It's not a crazy idea, but it's not obviously a panacea either.
All of which is to say, maybe there's something there... but often distributed databases have fewer features and have bigger footguns than rdbms's. If you want to try cockroach or aurora, you don't need the devs help. You can stay to stand up a Lemmy instance that points at them. If the compatibility is really good enough, it will "just work" and you can try some performance testing. If it doesn't, then you have your answer than porting effort is requires for speculative benefit.