Measuring DNS over TCP in the Era of Increasing DNS Response Sizes: A View from the Edge

Mike Kosek | Trinh Viet Doan | Simon Huber | Vaibhav Bajpai
Technical University of Munich

ACM SIGCOMM Computer Communication Review Volume 52 Issue 2, April 2022.

[Paper Arxiv →] https://arxiv.org/abs/2205.00746

[Paper ACM DL →] https://dl.acm.org/doi/10.1145/3544912.3544918

---

Reproducibility

In order to enable the reproduction of our findings, we make the raw data of our measurements as well as the analysis scripts and supplementary files publicly available within this repository.

Repository Overview

• analysis.ipynb is a jupyter notebook containing all analyses detailed in the paper
• public-resolvers-ipv4s.csv is single column text file containing a list of known public resolvers (used in related work)
• pyasn.dat is a 2 columns text file mapping RIPEAtlas (RA) probes IP address to the related ASN
• measurements.parquet contains the full measurements campaign run via RA probes

Each measurements sample has the following schema

field	example value	description
msm_id	29743869	Unique ID of the measurement
probe_id	21500	Unique ID of the RA probe
time	2021-04-19 14:58:56	Execution time of the measurement
proto	TCP	Transport protocol used for the measurement
src_address	192.168.111.130	Source IP address of the RA probe
dst_address	208.67.222.222	IP address of the targeted recursive DNS resolver
dst_port	53	Destination port of the measurement
result_rt	103.077	Response time of the measurement in ms
err_msg	None	Error reason if a measurement resulted in an error
edns_udp_size	4096	EDNS(0) buffersize signaled by the recursive DNS resolver
latitude	50.4975	Latitude of the RA probe
longitude	13.6275	Longitude of the RA probe
country_code	CZ	Country code of the RA probe
continent_code	EU	Continent code of the RA probe
resolver_name	OpenDNS	Name of the targeted Public DNS resolver, or Probe Resolver if not public
public_src_ip	213.129.132.83	Public IP address of the RA probe

Preparations

conda create -n ccr2022 pip
conda activate ccr2022
python -m pip install jupyterlab pandas matplotlib seaborn IPy pyasn pyarrow

Analysis

All results are generated from the notebook analysis.ipynb

---

Contact

Please feel welcome to contact the authors for further details.

• Mike Kosek (kosek@in.tum.de) (corresponding author)
• Trinh Viet Doan (doan@in.tum.de)
• Simon Huber
• Vaibhav Bajpai (bajpai@cispa.de)