https://github.com/LemmyNet/lemmy/issues/3245

I posted far more details on the issue then I am putting here-

But, just to bring some math in- with the current full-mesh federation model, assuming 10,000 instances-

That will require nearly 50 million connections.

Each comment. Each vote. Each post, will have to be sent 50 million seperate times.

In the purposed hub-spoke model, We can reduce that by over 99%, so that each post/vote/comment/etc, only has to be sent 10,000 times (plus n*(n-1)/2 times, where n = number of hub servers).

The current full mesh architecture will not scale. I predict, exponential growth will continue to occur.

Let’s work on a solution to this problem together.

  • HTTP_404_NotFound@lemmyonline.comOP
    link
    fedilink
    English
    arrow-up
    2
    ·
    edit-2
    1 year ago

    The centralized hub replication servers would have to basically foot a huge bill for the fediverse, and do so silently and invisibly to the end user.

    One consideration, since they are only having to basically sub/pub - the load actually might be drastically lower than expected.

    Furthermore the topology you propose is essentially what we already have. A few large instances hold most of the largest communities. I don’t see that changing.

    Suppose- that is a valid point. The issue though- those large instances are unable to keep up with demand and load, causing lots of federation issues.

    Perhaps, my idea actually wouldn’t help that at all, but, using lemmy.ml as an example-

    Instead of it having to send all of its updates out to every server subscribed- it can delegate that to a hub server to do it. The hub server can run a very minimal set of instructions, with enough intelligence to handle sub/pub.

    Perhaps- one idea is, instead of thinking of it as a hub-server, think of it as a proxy server. Being able to delegate your instances actions to the proxy server to reduce that load from the main server.

    And, instead of the hubs/proxies being more centralized, perhaps, its just an optional thing which you CAN do.

    My line of thinking, is methods to reduce load from the main servers. This might be an idea that only benefits the handful of big servers.

    • monobot@lemmy.ml
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      I am not certain on scenarios you were mentioning above, but I do agree that separating software to instance plus hub/proxym/mssage queue could help with handling load.

      How can we scale our big i instances? I don’t know maybe it is easy to put instance on multiple servers, but sounds to me they are just buying bigger one, and that will fill up fast of growth continues to happen.

      I would like to hear from developers what they think, but thank you for starting conversation about scaling.