USENIX Security2026

DaLens: Charting DNS Self-Amplification Threats at Large

Liwen Xu, Zechao Cai, Huayi Duan, Adrian Perrig

摘要

The emerging self-amplification attacks (SAAs) pose serious denial-of-service (DoS) risks to the Domain Name System (DNS). They can substantially amplify the interactions between recursive and authoritative servers, depleting resources at disproportionally small costs. Assessing the impact of such attacks on the global name resolution infrastructure is crucial for DNS operators to effectively triage threats and deploy defenses, yet this remains an uncharted and daunting territory. We have conducted the first large-scale measurement study of SAAs, leveraging a versatile framework, DaLens, which we designed and developed. The work consists of untangling the intricate ∈fra infrastructure to identify effective amplifiers and quantifying their amplification capabilities in a modular, scalable, and sound manner. Out of 307K persistent public resolvers, we find 29K unique resolver clusters that can be exploited in parallel for SAAs, and a significant number of them can still produce large amplification effects even though these vulnerabilities had already been disclosed in prior work.