* better error handling instead of just a stack trace, e.g. when alias doesn't exist
* add version selector to gem/package installs
* try to do less stuff as root
* extract ec2 specific code, and ubuntu specific code into a strategy/factory pattern so this can be used with other grid computing system
* figure out how to test the capistrano bits
* divorce from rails for use in deploying other things
* Script setup of round robin dns (Web role), mx records (mail role), etc in generators 
* add scripted failover to generated configs for:
    * remove web hosts from round robin on web host failure
    * db failover (promote slave replica? master-master replicas?  cluster?)