Whac-A-Mole: Six Years of DNS Spoofing

TL;DR

This paper analyzes six years of DNS data to detect and understand DNS spoofing, revealing its increasing prevalence and the mechanisms behind it, with implications for Internet security.

Contribution

It introduces methods to identify DNS spoofing, infer spoofing mechanisms, and analyze long-term trends using extensive data from root DNS servers.

Findings

DNS spoofing occurs in about 1.7% of observations

Spoofing has more than doubled over seven years

Proxies are the primary method of DNS spoofing

Abstract

DNS is important in nearly all interactions on the Internet. All large DNS operators use IP anycast, announcing servers in BGP from multiple physical locations to reduce client latency and provide capacity. However, DNS is easy to spoof: third parties intercept and respond to queries for benign or malicious purposes. Spoofing is of particular risk for services using anycast, since service is already announced from multiple origins. In this paper, we describe methods to identify DNS spoofing, infer the mechanism being used, and identify organizations that spoof from historical data. Our methods detect overt spoofing and some covertly-delayed answers, although a very diligent adversarial spoofer can hide. We use these methods to study more than six years of data about root DNS servers from thousands of vantage points. We show that spoofing today is rare, occurring only in about 1.7% of observations. However, the rate of DNS spoofing has more than doubled in less than seven years, and it occurs globally. Finally, we use data from B-Root DNS to validate our methods for spoof detection, showing a true positive rate over 0.96. B-Root confirms that spoofing occurs with both DNS injection and proxies, but proxies account for nearly all spoofing we see.