I have accounts on a few instances, and lemm.ee is the quickest and most stable of them all. I don't know what they're doing, but it's great.
Can't find the post right now but sunaurus said that the main difference is that lemm.ee is the only instance using a horizontal setup, that is, there's multiple lemmy servers running on multiple servers behind a load balancer, all sharing a database (postgres itself clusters very well). The code isn't actually made for that so it's all rather custom and possibly specific to his hoster.
Less technical: sunaurus happens to be a beast of a devop and prolific contributor to the lemmy codebase. As such lemm.ee quite often runs code that's ahead of the release schedule, addressing stuff that he stumbled across while wearing his sysop hat.
lemdro.id also runs via horizontal scaling behind a load-balancer, soon to expand globally to keep response times down for people everywhere. We're very resilient :)
I got accounts on lemm.ee, sh.itjust.works and kbin.social. I had one on .world in the beginning, but the performance wasn't great. Probably too many users.
Probably too many users.
if local.lemmyusers > 15, crash constantly because of PostgreSQL nonsense logic and Rust ORM.
That’s why I use lemm.ee
1993: God, how we would love it if someone could tell us anything was “just that simple”, and then of course when you see a pie chart you go “Oh, a pie chart…”. I mean, it has more religious meaning now than a crucifix to see a pie chart. I mean, because…. why is that so popular? Because it reduces complexity. The complexity is very real but his little soundbites - 1993
@garpunkal@lemm.ee - do you know of the history of site_aggregates PostgreSQL table?
huh?
huh?
Please explain in detail what "huh" means in this context.
As I said in the comment you replied to: do you know of the history of site_aggregates PostgreSQL table?
Not OP, but I feel like it was Huh? as in what the heck are you talking about and why was it a reply to thier comment
no tell me more?
lemmy.ca staff was so frustrated with performance problems a couple weekends ago they cloned a copy of their database Running AUTO_EXPLAIN revealed site_aggregates logic in Lemmy was doing comment = comment + 1 counting against 1500 rows, for every known Lemmy instance in the database, instead of just writing 1 row.
That's why I use lemm.ee
I have accounts on a few instances, and lemm.ee is the quickest and most stable of them all. I don't know what they're doing, but it's great.
Can't find the post right now but sunaurus said that the main difference is that lemm.ee is the only instance using a horizontal setup, that is, there's multiple lemmy servers running on multiple servers behind a load balancer, all sharing a database (postgres itself clusters very well). The code isn't actually made for that so it's all rather custom and possibly specific to his hoster.
Less technical: sunaurus happens to be a beast of a devop and prolific contributor to the lemmy codebase. As such lemm.ee quite often runs code that's ahead of the release schedule, addressing stuff that he stumbled across while wearing his sysop hat.
lemdro.id also runs via horizontal scaling behind a load-balancer, soon to expand globally to keep response times down for people everywhere. We're very resilient :)
I got accounts on lemm.ee, sh.itjust.works and kbin.social. I had one on .world in the beginning, but the performance wasn't great. Probably too many users.
if local.lemmyusers > 15, crash constantly because of PostgreSQL nonsense logic and Rust ORM.
1993: God, how we would love it if someone could tell us anything was “just that simple”, and then of course when you see a pie chart you go “Oh, a pie chart…”. I mean, it has more religious meaning now than a crucifix to see a pie chart. I mean, because…. why is that so popular? Because it reduces complexity. The complexity is very real but his little soundbites - 1993
@garpunkal@lemm.ee - do you know of the history of site_aggregates PostgreSQL table?
huh?
Please explain in detail what "huh" means in this context.
As I said in the comment you replied to: do you know of the history of site_aggregates PostgreSQL table?
Not OP, but I feel like it was Huh? as in what the heck are you talking about and why was it a reply to thier comment
no tell me more?
lemmy.ca staff was so frustrated with performance problems a couple weekends ago they cloned a copy of their database Running AUTO_EXPLAIN revealed site_aggregates logic in Lemmy was doing comment = comment + 1 counting against 1500 rows, for every known Lemmy instance in the database, instead of just writing 1 row.