TL;DR: DDIDD will apply existing and develop new defenses against Distributed-Denial-of-Service attacks
for operational DNS infrastructure, and make these tools available as open source.
Denial-of-Service (DoS) attacks and
Distributed-Denial-of-Service (DDoS) attacks are a continuing problem—attackers
Attackers employ spoofing, amplification,
and the use of very large botnets.
Their traffic can overwhelm even very well-provisioned services,
as shown by the huge Mirai attacks on Dyn in 2016.
DNS is a critical service on the Internet that many other services
depend on, and for DDIDD we are focused
on better securing DNS infrastructure against DDoS attacks.
DDIDD proposes to develop and deploy a defense-in-depth approach to mitigate
Distributed Denial-of-Service attacks for DNS servers.
Consistent with NSF’s goal for making Research cyber-infrastructure more resilient,
we seek to better protect operational DNS cyber infrastructure.
Our approach, Deep Layers,
will integrate approaches to filter spoofed traffic,
approaches to identify known-good traffic when possible,
and adds a cloud-based scaling component to handle the largest attacks.
These steps address an array of increasingly sophisticated attacks, ranging from those we
see today to those that may be possible in the future. In the end, we
hope to significantly increase the resilience of DNS
servers to DDoS attacks.
We plan to deploy Deep Layers to protect critical infrastructure services,
and to work with USC’s B-Root team as an initial case study.
We will be making our resulting tools available to others as open source software.
DDIDD builds on prior work from the STEEL lab
(FRADE and SENSS)
and USC’s B-Root.
DDIDD is supported by
Directorate of Computer and Information Science and Engineering (CISE)
as part of their
Cybersecurity Innovation for Cyberinfrastructure (CICI) program
as award number OAC-1739034.
Giovane C. M. Moura, John Heidemann, Wes Hardaker, Jeroen Bulten, Joao Ceron and Christian Hesselman 2020. Old but Gold: Prospecting TCP to Engineer DNS Anycast (extended). Technical Report ISI-TR-739b. USC/Information Sciences Institute.
ASM Rizvi, Joao Ceron, Leandro Bertholdo and John Heidemann 2020. Anycast Agility: Adaptive Routing to Manage DDoS. Technical Report arxiv:2006.14058v1. arXiv.
Lan Wei, Marcel Flores, Harkeerat Bedi and John Heidemann 2020. Bidirectional Anycast/Unicast Probing (BAUP):
Optimizing CDN Anycast. Proceedings of the IEEE Network Traffic Monitoring and Analysis Conference (Berlin, Germany, Jun. 2020).
John Heidemann, Wes Hardaker, Jelena Mirkovic, ASM Rizvi and Robert Story 2019. DDoS Defense in Depth for DNS (DDIDD). Invited talk at the Trusted CI Webinar.
ASM Rizvi, John Heidemann and Jelena Mirkovic 2019. Dynamically Selecting Defenses to DDoS for DNS (extended). Technical Report ISI-TR-736. USC/Information Sciences Institute.
Giovane C. M. Moura, John Heidemann, Ricardo de O. Schmidt and Wes Hardaker 2019. Cache Me If You Can: Effects of DNS Time-to-Live. Proceedings of the ACM Internet Measurement Conference (Amsterdam, the Netherlands, Oct. 2019), to appear.
Giovane C. M. Moura, John Heidemann, Moritz Müller, Ricardo de O. Schmidt and Marco Davids 2018. When the Dike Breaks: Dissecting DNS
Defenses During DDoS. Proceedings of the ACM Internet Measurement Conference (Oct. 2018).
Giovane C. M. Moura, John Heidemann, Ricardo de O. Schmidt and Wes Hardaker 2019. Cache Me If You Can: Effects of DNS Time-to-Live (extended). Technical Report ISI-TR-734b. USC/Information Sciences Institute.
For related publications, please see the
ANT publications web page.
The dnsroot filter for xtables filters for valid top level domains (TLDs).
See also the see the ANT distribution web page.
DDoS Filters for DNS Servers
We have developed and documented several DDoS defenses to assist during an attack on a DNS server:
In addition, we have experimented with the following filters (not yet released):
- in-kernel handling of NXDDOMAIN for non-TLDs with DPDK
- a known-host list that accepts traffic from prior known-good hosts while rejecting other traffic
We make all datasets and specifically
our network outage datasets
through the LACREND project.