Why do the instances keep going down? It makes me think that this is not a reliable social network, but the alternatives are not as good.

  • grte@lemmy.ca
    link
    fedilink
    arrow-up
    77
    ·
    1 year ago

    You’re thinking about it the wrong way. Despite a major hub of lemmy being down if you have an account on another instance you can continue using the network nearly as though nothing had happened. Individual instances may have greater or lesser reliability but the social network is very robust.

      • deweydecibel@lemmy.ml
        link
        fedilink
        arrow-up
        35
        arrow-down
        1
        ·
        edit-2
        1 year ago

        Then you should appreciate that the reliability of the social network is just fine. The idea is this social network isn’t dependent on one instance.

        Now, granted, if a big one struggles, the network loses some communities temporarily, but there the network is stable.

        It’s just growing pains from an extreme influx almost literally overnight and generally just that this is somewhat early days. It’s going to be messy. It always is, no matter what the social network.

        Also…there’s a non-zero chance it’s getting hit relentlessly by DDOS.

          • seang96@spgrn.com
            link
            fedilink
            arrow-up
            13
            ·
            1 year ago

            DDOS = denial of service attack. Attacker sends a bunch of requests overloading a service and causing other clients to experience.timeouts due to the service not being.abe.to.handle the load.

            • Konlanx@lemmy.ml
              link
              fedilink
              Deutsch
              arrow-up
              6
              arrow-down
              1
              ·
              1 year ago

              It’s like a group of people standing in line for the cashier and they each buy a single peanut with cash and have a question to the manager.

              I like that picture, it makes it easier to understand for people who aren’t that much into computers.

              • Jivebunny@lemmy.ml
                link
                fedilink
                arrow-up
                3
                ·
                1 year ago

                And now you can use that picture to even extend it with: We’re currently enjoying our checkout at different registers, where there’s not peanut nutjobs at the register. I like it too.

              • ndguardian@lemmy.studio
                link
                fedilink
                arrow-up
                2
                ·
                1 year ago

                Yep, this is key. If you’re getting a bunch of malicious traffic from one source, that’s easily fixed. Just drop the traffic.

                But when that traffic is coming from hundreds or thousands of sources, that becomes much harder to address. Can you just drop traffic from those sources? Sure! But then you also risk dropping legitimate traffic.

                There are also services that can automate the detection and prevention of DDOS attacks such as CloudFlare and Akamai, but these can get expensive very quickly, so it can significantly increase the cost to running the instance in question.

              • seang96@spgrn.com
                link
                fedilink
                arrow-up
                1
                ·
                1 year ago

                I honestly forgot what the first D was at that moment lol. While I agree it technically can be done pretty badly without distributed attacks. I read in the past couple of years of an approach attackers used was to make an application DOS itself from a single request. I think it required a vulnerability in the application in this instance though.

    • RoundSparrow@lemmy.ml
      link
      fedilink
      arrow-up
      3
      ·
      1 year ago

      You’re thinking about it the wrong way.

      I’ve had to go through a major change in thinking and adjust my interpretation in major ways.