Hacker News Reader

I wish that DNS could just switch over to another availability zone when this happens. A datacenter with all the sql replicated. Sure you'd pay twice as much for EBS, but it could also double as a backup. As for the other resources at the availability zone, they are hardly utilized until they spring into action (EC2, etc.)

View on HN

"...but it could also double as a backup."

Take care when treating a high-availability set up such as this as a back up: if you are replicating all the changes between 2 database servers and an application error (e.g. not a database outage) causes some kind of data corruption, you are hosed if the corruption replicates and you don't also have some previous "snapshot" of the data that you can roll back to.

You can do this, depending on your database. I wrote a blog post going over some of the techniques you can use with the AWS stack a while back - http://bit.ly/TD13iH

Reply on news.ycombinator.com