@viraptor thanks for the feedback. Excluding the "why" was a huge oversight. As far as running a local resolver, it depends on how you architect your infrastructure. Having a single line with 127.0.0.1 introduces a single point of failure. While that might be acceptable for demo/prototyping applications, I think for production deployments you'd want to avoid that. Also running a local resolver adds another service you'll need to monitor and keep up to date.<p>Note: This recommendation is for server setups, where dns latency can easily add ms to processing time. With workstations we can tolerate some latency and should probably benchmark your ISPs resolvers.