One of the first techniques to mitigate DDoS attacks is to minimize the surface area that can be attacked thereby limiting the options for attackers and allowing you to build protections in a single place. We want to ensure that we do not expose our application or resources to ports, protocols or applications from where they do not expect any communication. Thus, minimizing the possible points of attack and letting us concentrate our mitigation efforts. In some cases, you can do this by placing your computation resources behind Content Distribution Networks (CDNs) or Load Balancers and restricting direct Internet traffic to certain parts of your infrastructure like your database servers. In other cases, you can use firewalls or Access Control Lists (ACLs) to control what traffic reaches your applications.

The two key considerations for mitigating large scale volumetric DDoS attacks are bandwidth (or transit) capacity and server capacity to absorb and mitigate attacks.

Transit capacity. When architecting your applications, make sure your hosting provider provides ample redundant Internet connectivity that allows you to handle large volumes of traffic. Since the ultimate objective of DDoS attacks is to affect the availability of your resources/applications, you should locate them, not only close to your end users but also to large Internet exchanges which will give your users easy access to your application even during high volumes of traffic. Additionally, web applications can go a step further by employing Content Distribution Networks (CDNs) and smart DNS resolution services which provide an additional layer of network infrastructure for serving content and resolving DNS queries from locations that are often closer to your end users.

Server capacity. Most DDoS attacks are volumetric attacks that use up a lot of resources; it is, therefore, important that you can quickly scale up or down on your computation resources. You can either do this by running on larger computation resources or those with features like more extensive network interfaces or enhanced networking that support larger volumes. Additionally, it is also common to use load balancers to continually monitor and shift loads between resources to prevent overloading any one resource.

Whenever we detect elevated levels of traffic hitting a host, the very baseline is to be able only to accept as much traffic as our host can handle without affecting availability. This concept is called rate limiting. More advanced protection techniques can go one step further and intelligently only accept traffic that is legitimate by analyzing the individual packets themselves. To do this, you need to understand the characteristics of good traffic that the target usually receives and be able to compare each packet against this baseline.

A good practice is to use a Web Application Firewall (WAF) against attacks, such as SQL injection or cross-site request forgery, that attempt to exploit a vulnerability in your application itself. Additionally, due to the unique nature of these attacks, you should be able to easily create customized mitigations against illegitimate requests which could have characteristics like disguising as good traffic or coming from bad IPs, unexpected geographies, etc. At times it might also be helpful in mitigating attacks as they happen to get experienced support to study traffic patterns and create customized protections.