S3 Failure Raises Questions About Cloud Design

whydoeseverythingsuck.com

6 pointswhalliburton3 comments

Everyone is treating this failure as if them managing a MogileFS system (or just sticking the content on network attached storage) would never see downtime. It's silly.

Maybe you could have done better. Maybe you would have lost all your data permanently. Let's not be so over-the-top that we think our infrastructures don't have vulnerabilities. That's what I like about Smugmug. They at least cop to the fact that their more expensive internal systems had problems too.

I don't know the name for it, but there's a psychological difference between risks that you could have prevented and ones that you couldn't have. People think driving is safer than flying because they're doing the driving.

"But what we had never considered is that our "anti-normalization" design principle also relates to stability of design."

You might not have thought of it. But Werner Vogels did.

http://www.allthingsdistributed.com/2007/12/eventually_consi...

Reply on news.ycombinator.com