An architectural approach for mitigating next-generation denial of service attacks
MetadataShow full item record
It is well known that distributed denial of service attacks are a major threat to the Internet today. Surveys of network operators repeatedly show that the Internet's stakeholders are concerned, and the reasons for this are clear: the frequency, magnitude, and complexity of attacks are growing, and show no signs of slowing down. With the emergence of the Internet of Things, fifth-generation mobile networks, and IPv6, the Internet may soon be exposed to a new generation of sophisticated and powerful DDoS attacks. But how did we get here? In one view, the potency of DDoS attacks is owed to a set of underlying architectural issues at the heart of the Internet. Guiding principles such as simplicity, openness, and autonomy have driven the Internet to be tremendously successful, but have the side effects of making it difficult to verify source addresses, classify unwanted packets, and forge cooperation between networks to stop traffic. These architectural issues make mitigating DDoS attacks a costly, uphill battle for victims, who have been left without an adequate defense. Such a circumstance requires a solution that is aware of, and addresses, the architectural issues at play. Fueled by over 20 years worth of lessons learned from the industry and academic literature, Gatekeeper is a mitigation system that neutralizes the issues that make DDoS attacks so powerful. It does so by enforcing a connection-oriented network layer and by leveraging a global distribution of upstream vantage points. Gatekeeper further distinguishes itself from previous solutions because it circumvents the necessity of mutual deployment between networks, allowing deployers to reap the full benefits alone and on day one. Gatekeeper is an open-source, production-quality DDoS mitigation system. It is modular, scalable, and built using the latest advances in packet processing techniques. It implements the operational features required by today's network administrators, including support for bonded network devices, VLAN tagging, and control plane tools, and has been chosen for deployment by multiple networks. However, an effective Gatekeeper deployment can only be achieved by writing and enforcing fine-grained and accurate network policies. While the basic function of such policies is to simply govern the sending ability of clients, Gatekeeper is capable of much more: multiple bandwidth limits, punishing flows for misbehavior, attack detection via machine learning, and the flexibility to support new protocols. Therefore, we provide a view into the richness and power of Gatekeeper policies in the form of a policy toolkit for network operators. Finally, we must look to the future, and prepare for a potential next generation of powerful and costly DDoS attacks to grace our infrastructure. In particular, link flooding attacks such as Crossfire use massive, distributed sets of bots with low-rate, legitimate-looking traffic to attack upstream links outside of the victim's control. A new generation of these attacks could soon be realized as IoT devices, 5G networks, and IPv6 simultaneously enter the network landscape. Gatekeeper is able to hinder the architectural advantages that fuel link flooding attacks, bounding their effectiveness.
RightsAttribution-ShareAlike 4.0 International