Abstract

The Domain Name System (DNS) constitutes an essential component of Internet infrastructure serving as a translator of human-readable domain names into machine-readable IP addresses. Although simple in its operation, DNS performance, especially its responsiveness, has a bearing on the user experience, web application performance, and network resilience. This research examines the responsiveness of DNS across several top-level domains (TLDs) including .com, .org, .net, .edu and the newer generic TLDs. tech and. xyz. Using a set of recursive DNS resolvers distributed globally, this research resolves TLDs in parallel over long durations to capture rich datasets. Systematic assessment is performed on response latency due to propounded factors such as TTLs, cache performance, the location of authoritative servers, and the efficiency of anycast routing. The employed techniques of real-time probing, latency aggregation, temporal sampling, and statistical smoothing aid in consistency, mitigating transient network disturbances. A hybrid analytical model of time-series analysis and decision-tree based classification is applied to cluster domains revealing performance profile similarities. The legacy TLDs were observed to outperform newer counterparts, sustaining the argument about the disparity of response times across TLDs, newer TLDs were found to lag due to immature infrastructure and poorly distributed name servers. These findings highlight DNS optimization needs, particularly for emerging TLDs to improve user experience and site responsiveness. As a practical reference for network architects and domain registrars, the paper provides a scoring framework for assessing the performance of a DNS at the TLD level.