At the time of writing, Lemmyworld has the second highest number of active users (compared to all lemmy instances)

Also at the time of writing, Lemmyworld has >99% uptime.

By comparison, other lemmy instances with as many users as Lemmyworld keep going down.

What optimizations has Lemmyworld made to their hosting configuration that has made it more resilient than other instances’ hosting configurations?

See also Does Lemmy cache the frontpage by default (read-only)? on !lemmy_support@lemmy.ml

  • Ruud@lemmy.worldM
    link
    fedilink
    arrow-up
    2
    ·
    1 year ago

    Thank you for all the compliments.

    This ride reminds me of Mastodon.world in November. Details on that are here: https://blog.mastodon.world/and-then-november-happened

    So I started lemmy.world on a 2CPU/4GB VPS. Keeping an eye on the performance. Soon I decided to double that. And after the first few thousand of users joined, doubled it again to 8CPU/16GB. That also was the max I could for that VPS type.

    But, already I saw some donations come in, without really asking. That reminded me of the willingness to donate on Mastodon, which allowed me to easily pay for a very powerful server for mastodon.world, one of the reasons it grew so fast. Other (large) servers crashed and closed registrations, I (mainly) didn’t.

    So, I decided to buy the same large server (32cpu/64threads with 128GB RAM) as for masto (but that masto one has double the RAM). With the post announcing that, I also mentioned the donation possibilities. That brought a lot of donations immediately, already funding this server for at least 2 months. (To the anonymous person donating $100 : wow!).

    Now next: to solve the issue with post slowness. That’s probably a database issue.

    And again: migration took 4 minutes downtime, and that could have been less if I wasn’t eating pizza at the same time. So if any server wants to migrate: please do! If you have the userbase, you’ll get the donations for it. Contact me if you have questions.

    • Xaphanos@lemmy.world
      link
      fedilink
      arrow-up
      1
      ·
      1 year ago

      As someone “in the business”, but not nearly as technical as you… How far can a single instance scale? Can a load balancer spread it over mulitple front-ends to handle user load? Can the back-end be re-worked to handle hundreds of millions of user operations per second? Can it work with a CDN? Can a single “Lemmy.World” site exist as a distributed site - with hundreds of servers spread across dozens of sites across the globe?

      I expect this entire line of thought is antithetical to the entire Lemmy philosophy of distributed operation. I expect that the “correct” way is to spin off “NA.Lemmy.World”, EMEA. Lemmy.World", APAC.Lemmy.World", etc. as separate servers. Is that correct?

      Thanks.

    • million@lemmy.world
      link
      fedilink
      arrow-up
      1
      ·
      1 year ago

      Nice job, thanks very much for the write up.

      Out of curiosity are you cloud hosting or do you own a server on a rack somewhere? Scaling with Kubernetes or VMs or just running bare-metal?