AS216394 · WireShield

Attack Analysis

What is a DDoS attack? How it works and how to protect against it

A DDoS attack makes a service unreachable by flooding it with traffic from thousands of devices. Learn how it works, the main attack types, warning signs and protection options.

What is a DDoS attack: attack traffic from many devices overloads a single server

A DDoS (distributed denial-of-service) attack tries to make a system or network unreachable for real users by overwhelming it with traffic from many devices at once. The US cybersecurity agency CISA describes denial-of-service attacks as a malicious actor preventing legitimate users from accessing an information system. This guide answers what a DDoS attack is, how it works, how to recognise one and how to protect against it.

What is a DDoS attack?

The goal of a DDoS attack is to exceed the capacity of a server, an application or an internet connection. Capacity can run out in three places: the link's bandwidth, the packet-processing power of network devices and servers, or the request-handling capacity of the application. Once one of them is full, legitimate visitors cannot reach the service either.

"Distributed" means the traffic comes from many devices on many networks rather than from one source. That makes defence harder, because blocking a single IP address does not stop the attack. We cover the difference to a single-source attack in DDoS vs DoS.

How does a DDoS attack work?

Most DDoS attacks follow the same three steps:

  1. Recruiting devices: The attacker takes over computers, cameras and routers that have vulnerabilities or default passwords. Together these devices form a botnet. See what is a botnet for details.
  2. Command: The attacker tells the botnet which target to hit and with which method.
  3. Traffic: The devices start sending requests, packets or connections to the target at the same time.

Some attacks do not even need the attacker's own devices. In reflection and amplification, the attacker sends small requests with the victim's address as the spoofed source to open DNS, NTP or memcached servers, which then send much larger responses to the victim. According to the table published by CISA, the amplification factor is 556.9 for NTP, 28 to 54 for DNS and 10,000 to 51,000 for memcached.

Types of DDoS attacks

Attacks fall into three groups depending on what they exhaust:

GroupWhat it exhaustsExamples
VolumetricLink bandwidthUDP flood, ICMP flood, DNS and NTP amplification
ProtocolConnection and packet-processing capacity of devicesSYN flood, ACK/RST flood, fragmented packet attacks
Application layerRequest capacity of the web server or applicationHTTP flood, slow request attacks, HTTP/2 Rapid Reset

We explain how each type works, how it shows up and how it is defended in types of DDoS attacks. In practice, attackers often combine several methods at once; this is called a multi-vector attack.

Two real-world examples

  • GitHub, February 28, 2018: According to GitHub's own incident report, the site was hit by a memcached-based UDP amplification attack that peaked at 126.9 million packets per second and 1.35 Tbps. The site was unavailable from 17:21 to 17:26 UTC and intermittently affected until 17:30.
  • HTTP/2 Rapid Reset, 2023: According to Google's October 10, 2023 disclosure, attacks that peaked in August reached more than 398 million requests per second, the largest attack of its kind at the time. The flaw is tracked as CVE-2023-44487. It shows that an attack can target the application layer instead of bandwidth.

How do you know you are under a DDoS attack?

None of these signs is proof on its own, but together they point to an attack:

  • A sudden, unexplained traffic spike, especially to one port or with one type of packet.
  • Slow responses, timeouts and connection errors.
  • Requests for the same page from many countries or networks at once.
  • CPU, connection table or bandwidth usage hitting the ceiling quickly.

We describe which data to watch in the UDP flood monitoring guide and how to read the result in reading an attack report.

How do you protect against DDoS attacks?

Effective protection comes from layers, not a single tool:

  1. Filter in front of your link. If attack traffic has already filled your link, a firewall on the server is too late. Remote DDoS protection routes traffic over BGP to a filtering network and returns only the clean part to you.
  2. Filter inside your network too. For attacks that do not fill the link, inline protection with WireShield Edge works on site. We compare the models in inline or remote.
  3. Protect web applications separately. Application-layer attacks such as HTTP floods are handled by Web Protection with a WAF, rate limiting and challenge pages.
  4. Close open services. An open DNS resolver, NTP or memcached server can be abused against others. Close or restrict services you do not need.
  5. Plan ahead. Decide who does what when an attack starts, and have your prefix list, ASN and contacts ready.

Common mistakes

  • Expecting the server firewall to absorb the attack.
  • Blocking one IP address and assuming the problem is solved.
  • Protecting only the network layer and leaving the web application exposed.
  • Looking for protection after the attack has started.

Frequently asked questions

Is a DDoS attack illegal?

Yes. Launching a denial-of-service attack against a system without permission is a crime in most countries. Using services sold as "stressers" or "booters" falls under the same rule.

Does a DDoS attack steal data?

Usually not. The goal is to interrupt the service, not to take data. However, some attacks are used as a smokescreen to distract the security team from another intrusion.

Do small businesses get attacked too?

Yes. Attacks do not only target large companies; small services are hit by competitors, for revenge or for extortion. Automated tools keep the cost of an attack low.

How long does a DDoS attack last?

From minutes to days. A single intense wave can last a few minutes, as in the GitHub case, while other attacks come back in waves over a day. Duration depends on the attacker's goal and the effectiveness of the defence.

Does a firewall stop DDoS?

A network or server firewall alone cannot stop a volumetric attack, because the link is full before the traffic reaches the firewall. That is why filtering has to happen in front of the link.

Sources

Next step

Let us assess together how much risk you have at each layer. Describe your network, prefixes and web applications, and we will suggest the right protection model. You can talk to the technical team.

All posts

Let’s review the right protection model for your network together.

Contact our team to talk through your technical needs.