• melroy@kbin.melroy.org
    link
    fedilink
    arrow-up
    53
    ·
    5 months ago

    The article says; if they self host it will cost them billions of dollars.

    But I don’t believe that at all. In fact, self hosting can be much cheaper on the long run.

    This is the reason Bluesky apparently can scale so well, they use their own infra. Hack, I’m now sending this message from my own infra

    • Alvaro
      link
      fedilink
      arrow-up
      1
      ·
      6 hours ago

      The issue, especially with real-time communications, is that you need infrastructure in global positions. They need to be close to people, they need to have high bandwidth. These are things that are really complicated both financially and legally. You have to have legal representatives in all kinds of countries, so it is much more than just the infrastructure, which is the servers and the data centers and that is why AWS and other cloud providers do make a lot of sense for small or low revenue companies.

      Self-hosting really makes sense when you’re very small or very large but the intermediate position does make a lot of sense to use cloud providers and that is why this model did succeed.

    • TrickDacy@lemmy.world
      link
      fedilink
      arrow-up
      18
      ·
      5 months ago

      You’re saying a single company can buy and maintain a server infrastructure cheaper than rates like .0001 cent per request? Yeah I don’t quite believe that. An entire industry moved to using AWS because it was cheaper.

      AWS sucks for several reasons but let’s not pretend it’s more expensive than self hosting

        • melroy@kbin.melroy.org
          link
          fedilink
          arrow-up
          3
          ·
          5 months ago

          its cost more money upfront, since companies need to invest money to build their servers/server racks. You can also still rent space in a data-center, without the need of building your own data center.

          But on the long run, it can be much cheaper than constantly renting all the hardware. You can compare it to houses, buying a house costs more money then renting. But overall in the long run, you are normally better off buying a property (assuming you can of course… its just an example).

        • IphtashuFitz@lemmy.world
          link
          fedilink
          English
          arrow-up
          2
          ·
          5 months ago

          The issue with cloud providers like AWS is that they charge for virtually everything, and that makes it easy to rack up charges if you forget about something you spun up as a test last week and forgot to terminate it. For larger companies it can be a significant issue. So there are other companies out there that you can use to scan your entire AWS account, summarize what you’re using, and highlight things you may not need any more. They’ll also recommend cost savings measures like paying for a year of server time up front instead of paying as you go. If you know you’ll need a server for a year then paying annually is a lot less expensive.

          On the plus side, you don’t need to deal with things like hardware failures. We have a large AWS environment where I work, and we’ll occasionally get an email informing us that an instance is “running on degraded hardware”. A simple reboot (power cycle) will move the instance to new hardware. And if you decide you need more RAM, more CPUs etc. then it’s also as simple as rebooting.

        • TrickDacy@lemmy.world
          link
          fedilink
          arrow-up
          1
          ·
          5 months ago

          Their servers are slow, I have seen that myself, but I don’t see how it wouldn’t be cheaper to use AWS other than maybe some highly specific scenarios.

          • Count042@lemmy.ml
            link
            fedilink
            English
            arrow-up
            2
            ·
            5 months ago

            You have it backwards.

            There are some very few specific use case that most companies don’t ever meet that makes AWS cheaper. In the vast majority of use cases it is an order of magnitude more expensive.

      • melroy@kbin.melroy.org
        link
        fedilink
        arrow-up
        1
        ·
        5 months ago

        Signal Foundation is indeed non profit… That being said OpenAI used to be non profit as well hahaha. And yes Bluesky is for-profit, just like X, Facebook etc.

    • Gelik@feddit.dk
      link
      fedilink
      arrow-up
      13
      ·
      5 months ago

      Yeah, for example Microsoft Azure and Google’s cloud. They operate on a global scale too

    • Dragonstaff@leminal.space
      link
      fedilink
      English
      arrow-up
      7
      ·
      5 months ago

      The question isn’t “why does Signal use AWS?” It’s to look at the infrastructural requirements of any global, real-time, mass comms platform and ask how it is that we got to a place where there’s no realistic alternative to AWS and the other hyperscalers. 3/

      https://bsky.app/profile/meredithmeredith.bsky.social/post/3m46a2fm5ac23

      She was misquoted (although the meaning should have been clear). This isn’t just “cloud” and bears no resemblance to a web server you spun up at home. This sort of world spanning tech stack is not something any company can build themselves, and there are only 3 or 4 companies that could host Signal.

      The world’s Internet infrastructure basically supports civilization as we know it, and it’s crazy to allow it to be privately owned with so little competition.

      In the old days, there would be public standards and interoperability and networks of organizations working together. Now the Internet is a series of proprietary walled gardens.

      • quick_snail@feddit.nl
        link
        fedilink
        arrow-up
        3
        ·
        5 months ago

        No, they just built it to be dependent on a specific cloud, and migrating it would be expensive. Due to bad decisions

      • quick_snail@feddit.nl
        link
        fedilink
        arrow-up
        4
        ·
        5 months ago

        Would be nice if the EU ran free matrix servers for their citizens.

        Germany already runs mastodon for their government ministries.

      • quick_snail@feddit.nl
        link
        fedilink
        arrow-up
        4
        ·
        5 months ago

        In a federated infrastructure, the answer is “any or all governments”

        Tax dollars support devs who submit PRs and hosting server instances

    • Ferk@lemmy.ml
      link
      fedilink
      arrow-up
      14
      ·
      5 months ago

      Or distributed serverless P2P communication (like SimpleX does). Specially when it comes to an app that is just meant for person-to-person communications to begin with.

      • als
        link
        fedilink
        English
        arrow-up
        2
        ·
        5 months ago

        SimpleX have message relay servers that are required for the sytem to function. It’s not “serverless P2P”.

    • Not_mikey@lemmy.dbzer0.com
      link
      fedilink
      arrow-up
      2
      ·
      5 months ago

      You can run your own signal server and federate it with others, you just can’t on the standard app you get from the app store that just talks to the central signal server.

      It’s all open source though so you’d just need to flip some conf flags and compile it yourself.

  • shortwavesurfer@lemmy.zip
    link
    fedilink
    arrow-up
    15
    ·
    5 months ago

    I’m going to call bullshit. There are several decentralized storage networks and resource allocation networks over blockchains.

    • melroy@kbin.melroy.org
      link
      fedilink
      arrow-up
      13
      ·
      5 months ago

      You don’t need block chain. They just can start to self host, instead of joining aws like every other company.

      No sht that we only have 4 large cloud providers, it’s because all there customers are lazy and do not want to self host.

  • Ŝan • 𐑖ƨɤ@piefed.zip
    link
    fedilink
    English
    arrow-up
    7
    ·
    5 months ago

    For þem and þeir architecture, probably. Þat says more about þe quality of þeir systems design, þan anyþing else.

      • Geodad@lemmy.world
        link
        fedilink
        arrow-up
        3
        ·
        5 months ago

        It doesn’t rely on Amazon not fucking up DNS traffic, and I control it because it’s my hardware.

        Every Signal video call I have ever been a part of has had shit for both audio and video quality. It’s not a hardware issue because everyone involved has flagship model phones.

        Signal has it’s use as an encrypted text message alternative.

        • notarobot@lemmy.zip
          link
          fedilink
          arrow-up
          2
          ·
          5 months ago

          OK, cool. So the answer is no then? you didn’t really answer.

          At some point you are relying on someone not fucking up something somewhere. At the very least you need your ISP not fucking up your connection speed or something similar.

          I’m not saying that xmpp sucks or that they are right on saying that there are not alternatives (although I am inclined to agree). What I’m saying is that your server is not a reference point to compare against, because you operate at immensely different scales and requirements

  • fruitycoder@sh.itjust.works
    link
    fedilink
    arrow-up
    1
    ·
    5 months ago

    Hybrid multi cloud is what every mature org moves too…

    Like eventually you just cant justify being on only one cloud (businesses, cost and administrative risks), and if you have a consistent enough usage scaling into the cloud for the baseline is just an unjustifiable expense