Learning from the S3 Outage

Today’s S3 outage taught me quite a bit. I don’t currently have any sites or services running on AWS, but I will have multiple ones up shortly. I am always big on redundancy, but I admit that when I comes to things like EC2 and S3 – I assumed that Amazon had already implemented the redundancy (trust me, I realize that this wasn’t the best viewpoint to take). When it boils down to it – here is the truth: Everything can fail. Second, if it isn’t your service/server/network – you can’t fix it. You just simply have to wait for it to be fixed.

Every site needs a ‘what if’ plan: what if my database server crashes, what if S3 goes down, what if EC2 goes down and I lose all of my server instances? In a world that is moving so fast toward hosted services – we need to remember these things and architect our applications to function (even if not at a full level) when these things fail. Oh, and don’t kid yourself – they will fail.