: Fault in DynamoDB system cascaded through AWS services, knocking major sites offline for hours
      • eldavi@lemmy.mlEnglish
        6·
        11 hours ago

        is the point of this comic to express how a single point of failure can bring a system down and the dns here definitely feels like a single point.

    • Matt The Horwood@lemmy.horwood.cloudOPEnglish
      31·
      15 hours ago

      not sure, it looks like its a different system that manages the DNS for DynamoDB. it got out of sync and deleted all the DynamoDB DNS records, that then broke all the internal AWS stuff

      • azimir@lemmy.ml
        3·
        7 hours ago

        AWS is mostly built on AWS.

        Yes, DNS started failing to properly fill name lookups. So, DynamoDB started failing. That started making security and other AWS services fail. Which in turn made higher level services fail.

        It truly was a house of cards kind of moment.

        • Admetus@sopuli.xyz
          1·
          5 hours ago

          Reading that article as a layman, I got the impression there were a lot of falling dominoes.

          • JGrffn@lemmy.ml
            4·
            4 hours ago

            Unexpected behavior, leading to unexpected inputs to systems, often leads to failures. Put simply, nobody thought X would happen on Y service or at Z level, so nobody wrote code to handle that scenario. May sound crazy at first, but it’s quite hard to cover all possible scenarios when writing code… Or with life in general.