newbie question: I understand that by delegating dns resolution for domains (and subdomains) to their zones, dns scales well. However, isn't there a huge load on the global top level domains.
I imagine most sites use .com and all dns requests for www.xxx.com , the queries for . and .com may be heavily cached, but if xxx isn't very popular , would that not put a huge number of requests against the .com nameservers to resolve to the xxx.com domain?If we had a huge number of low popularity xxx domains in the com domain, won't all dns queries have to hit the top level .com nameservers?
Yes. There are currently 13 virtual nodes that handle .com and, just like for the root servers, each one of those virtual nodes is actually a group of machines all bound to the same IP address via Anycast. There are hundreds of physical machines spread around the world handling the TLDs. The NS records that these servers return are readily cached with long lifetimes, though, so if you use any relatively popular DNS server, either your ISPs or Google's public server, the resolution process will be really short even for uncommon names.
