Hi, I'm new...
I'm a webmaster trying to understand DNS. I'm using loadbalancing with failover for my website and I'm trying to establish a small network of hosts near to my main sources of traffic.
I'm inclined to think that clients will connect to the host that is nearest in the network, but my computer invariably connects to the host farthest away, even after running ipconfig/flushdns. My domain's A records have 300s TTL.
But then again, if I use traceroute to my domain, from various locations in the world, then the nearest host is always chosen.
My website's access logs don't suggest that people get connected to nearby hosts, but rather that it's random.
1. One of the things I'd like to know is: if a person on a PC tries to connect to a domain with A records pointing to multiple hosts, what picks the host? The PC's OS, the ISP or the DNS server?
2. And why is there a difference in behaviour between my PC and traceroute utilities and why do my site's access logs suggest randomness if the former two don't?
If anyone can answer these questions, then that would great, or else links to some literature on this will also be appreciated.
Thanks!
Martin