Currently we have on c1.medium webnode and an Amazon RDS sql server. The problem we face is that every now and then the webnode fails and the site goes down. I am considering mitigating this risk by running two or three webnodes behind an ELB. However, the cost of doing this is discouraging, as we have already purchased reserved instances for our current set up.
My questions are:
- How does ELB distribute traffic amongst the nodes behind it? By response time? By location?
- Can I get AWS to restart dead nodes automatically? Where does the code that does this run? On existing healthy web nodes?
- Will I need to set up something like GlusterFS for this configuration?