Good morning,
I am currently analysing the DNS resolvers (local and public ones) in terms of protection and performance (in particular their speed).
I noticed that, in case of a malicious domain name, some local resolvers send an NXDOMAIN and others a courtesy page address. Do you know if the resolvers (for example TIM, Wind or Fastweb) can return an NXDomain in order to protect their clients?
From a network engineering perspective, any resolver that responds to an authoritative NXDOMAIN by generating an address for a courtesy page -is- the malicious actor. Doubly so if they lie about the DNSSEC status in the response.
Nevermind; I misunderstood your question. The domain name exists but
the resolver has blocked it. How should the resolver alter its
response: NXDOMAIN or the IP address of a courtesy web page explaining
the block.
Resolvers are capable of rewriting a response to anything they want. In the case of filtering out known bad networks, you can find examples of both rewriting to a courtesy web page and NXDOMAIN. There is a scheme known as Response Policy Zone1 that hasn’t been standardized (yet?) but is available in some recursive DNS software, such as BIND, which lets you do either.
As for which large operators respond in different ways, I’m afraid I can’t help you there. I’m not aware of any surveys done of how individual large operators implement their end user protection services.