It seems like when something does break, ebs failures cascade to other services (RDS) and you're pretty much hosed... Ever since the first big us-east-1 ebs ou

It seems like when something does break, ebs failures cascade to other services (RDS) and you're pretty much hosed... Ever since the first big us-east-1 ebs outage I've tried to avoid ebs, but I am using it for root volumes on most of my instances which makes me nervous -- so many instance types don't have instance storage, I guess I need to look at c5d instances again. I thought this writeup my smugmug was pretty good and still applies: https://don.blogs.smugmug.com/2011/04/24/how-smugmug-survived-the-amazonpocalypse/ https://don.blogs.smugmug.com/2011/04/24/how-smugmug-survive...