RightScale Blog

Cloud Management Blog
Cloud Management Blog

Add new comment

Hi Jon, I certainly agree with you on the sub-optimal client experience and I don't think that the goal is to stay in the 'failed' state for a long period of time, rather to keep the site/service up and running while recovery is underway. I also worked this out using an existing DNS provider-I generally don't set up DNS load balancers for cloud deployments, but it's definitely something worth checking out. My main concern about a single 'A' record is that once that server fails, there needs to be a plan to manage the failure of the DNS load balancer, so the other goal of two ingress IP addresses (in this case) was to mitigate the risk of either one of them going down. Setting up a redundant DNS infrastructure across disparate zones/cloud regions or providers could also be an option. Best, Patrick
Posted by Patrick McClory   Ι   October 24, 2012   Ι   08:51 AM