I call it bullshit too. If its too expensive for them, just decentralize the project. Self-hosters all around the world would help. I alone have better uptime than AWS and probably wouldn’t even notice usage from a few hundred thousands users
you, on a single ISP who relies on the world’s shared backbone rather than your own between multiple DCs within a region and multiple regions around the world, have better uptime than AWS?
Stop.
I’m all for decentralizing for the case of no single entity controlling everything, but not for the case of uptime. That is one thing you give up with services like Matrix or Lemmy.
AWS actually has an SLA it’s contractually committed to when you pay them with thousands of engineers working to maintain it.
Well yes considering the downtime they had. SLA is just words on a paper, you also need to not fuck your infrastructure up. Even if all self-hosters had 99% uptime which is bad, it’s easy building a system that replicates data on a few of them to achieve resiliency. People need to stop assuming they can be 100% reliant on a single host and actually design their systems to take downtimes into account and recover from them
I call it bullshit too. If its too expensive for them, just decentralize the project. Self-hosters all around the world would help. I alone have better uptime than AWS and probably wouldn’t even notice usage from a few hundred thousands users
You cant run a professional service on self hosters hardware…
I think you guys dont really have experience of building these global, low latency apps and dont know the challanges that come with that…
you, on a single ISP who relies on the world’s shared backbone rather than your own between multiple DCs within a region and multiple regions around the world, have better uptime than AWS?
Stop.
I’m all for decentralizing for the case of no single entity controlling everything, but not for the case of uptime. That is one thing you give up with services like Matrix or Lemmy.
AWS actually has an SLA it’s contractually committed to when you pay them with thousands of engineers working to maintain it.
Well yes considering the downtime they had. SLA is just words on a paper, you also need to not fuck your infrastructure up. Even if all self-hosters had 99% uptime which is bad, it’s easy building a system that replicates data on a few of them to achieve resiliency. People need to stop assuming they can be 100% reliant on a single host and actually design their systems to take downtimes into account and recover from them
It’s not just words on paper. It’s a level of service you commit to and owe repercussions when broken.
You get compensated when broken, but it’s still broken