Richard Hurt

"Well, it depends on your mind-set and how many servers you're managing. If you're only baby-sitting 2-3 servers, then OK I can see where you might have a login for testing, building configurations, etc. When you're managing hundred's of servers, which is where cloud computing shines, then things are different. Doing things the "cloud" way means that you never log into a server no matter what, you just don't have time. If there is a problem with a process, kill the server and spawn another one. If you want to watch a process to make sure its working correctly, use monitoring software (CloudWatch). If you need to know more about processes, log the data somewhere and investigate it." - Richard Hurt

"Chef was a bit confusing for me at first but I'm getting used to it and even enjoying it a little bit. I found that Amazon's new (free) OpsWorks service was a pretty good introduction to Chef for me. Basically, OpsWorks forces you into doing the "right thing" by using all the things I mentioned." - Richard Hurt

"Generally, you shouldn't use EC2 instances as "permanent" servers. The best way to use cloud resources is to treat them as transitory and plan for them to fail at any time. Use tools like Chef/Puppet to store your configurations, S3 to store persistent data, and Github to store source code. This allows you to recreate servers and even duplicate servers (failover) with minimal pain and suffering. I would even go so far as making it so that you cannot SSH into the server itself. If you're logging in to make changes or check processes, you're doing it wrong." - Richard Hurt

"Good [picture](http://i.dailymail.co.uk/i...) showing the webbing between the eye and nose." - Richard Hurt