The Domain Name System is one of the fundamental services of the Internet, but its original design does not support cryptographic verification of the authenticity and integrity of DNS data, nor does it provide confidentiality of DNS queries. Consequently, DNS is exposed to various security risks, including response spoofing, cache poisoning, traffic interception and modification, and reflection-amplification attacks.
This thesis examines the security mechanisms DNSSEC, DNS-over-TLS and DNS-over-HTTPS, as well as their roles in securing the DNS. The practical part was conducted using two complementary test scenarios. A BIND9 authoritative server and a Kali Linux client system were set up within a controlled Oracle VirtualBox environment. Implementations included automatic zone signing with DNSSEC, validation using a local trust anchor, DoT via an NGINX proxy, and DoH using dnsdist software. The solutions were verified through functional tests, a negative DNSSEC test, and network traffic analysis using Wireshark.
The experiment was supplemented by measurements taken over an actual Internet connection to a public recursive resolver. One hundred measurements were performed for each protocol (DNS/UDP, DNS/TCP, DoT and DoH) using the same client tool. The average response times were 29.8 ms for DNS/UDP, 41.6 ms for DNS/TCP, 250.9 ms for DoT, and 251.1 ms for DoH. With DoT and DoH, a new encrypted connection was established for each query. Consequently, the results include the overhead of establishing the TCP/TLS or HTTPS session and cannot be equated to the performance of long-lived connections.
Practical tests have confirmed that DNSSEC, DoT and DoH address different security issues. DNSSEC enables the verification of origin authenticity and data integrity for DNS but does not ensure query confidentiality. In contrast, DoT and DoH encrypt the communication channel between the client and the resolver, though they do not inherently provide DNS data validation. Therefore, a combination of cryptographic validation, encrypted transport, and additional operational mechanisms is advisable for more comprehensive DNS system protection.
|