The scourge of distributed denial-of-service (DDoS) attacks has been a major concern for businesses and governments for more than two decades. First reported in 1996, this is a destructive and ever-evolving vector of cyber raids that knocks electronic networks offline by flooding them with the traffic they can’t handle.
Not only is DDoS a way for hacktivists to manifest protest against Internet censorship and controversial political initiatives, but it’s also a goldmine of opportunities for achieving strictly nefarious goals. For instance, the latest tweak in this epidemic is what’s called “ransom DDoS,” a technique used to extort money from organizations in exchange for discontinuing a massive incursion.
A big hurdle to thwarting the DDoS phenomenon is that it’s heterogeneous and spans a variety of different tactics. To begin with, there are three overarching categories of these attacks that form the backbone of this ecosystem:
- Volume-based (volumetric) attacks are the “classic” ones that congest a target network’s bandwidth with a hefty amount of traffic packets.
- Protocol attacks are aimed at exhausting server or firewall resources.
- Application layer (layer 7 DDoS) attacks zero in on specific web applications rather than the whole network. These ones are particularly hard to prevent and mitigate while being relatively easy to orchestrate.
Furthermore, there are dozens of sub-types that fall into either one of the above generic groups but exhibit unique characteristics. Here’s a complete breakdown of the present-day DDoS attack methods.
This attack exploits the TCP three-way handshake, a technique used to establish any connection between a client, a host, and a server using the TCP protocol. Normally, a client submits a SYN (synchronize) message to the server to request a connection.
When a SYN Flood attack is underway, criminals send a plethora of these messages from a spoofed IP address. As a result, the receiving server becomes incapable of processing and storing so many SYN packets and denies service to real clients.
To perform a Local Area Network Denial (LAND) attack, a threat actor sends a fabricated SYN message in which the source and destination IP addresses are the same. When the server tries to respond to this message, it gets into a loop by recurrently generating replies to itself. This leads to an error scenario, and the target host may eventually crash.
The logic of this attack vector is to abuse the TCP communication stage where the server generates a SYN-ACK packet to acknowledge the client’s request. To execute this onslaught, crooks inundate the CPU and RAM resources of the server with a bevy of rogue SYN-ACK packets.
ACK & PUSH ACK Flood
Once the TCP three-way handshake has resulted in establishing a connection between a host and a client, ACK or PUSH ACK packets are sent back and forth until the session is terminated. A server targeted by this type of a DDoS attack cannot identify the origin of falsified packets and wastes all of its processing capacity trying to determine how to handle them.
Fragmented ACK Flood
This attack is a knockoff of the above-mentioned ACK & PUSH ACK Flood technique. It boils down to deluging a target network with a comparatively small number of fragmented ACK packets that have a maximum allowed size, usually 1500 bytes each. Network equipment such as routers ends up running out of resources trying to reassemble these packets. Furthermore, fragmented packets can slip below the radar of intrusion prevention systems (IPS) and firewalls.
Spoofed Session Flood (Fake Session Attack)
In order to circumvent network protection tools, cybercriminals may forge a TCP session more efficiently by submitting a bogus SYN packet, a series of ACK packets, and at least one RST (reset) or FIN (connection termination) packet. This tactic allows crooks to get around defenses that only keep tabs on incoming traffic rather than analyzing return traffic.
As the name suggests, this DDoS attack leverages multiple User Datagram Protocol (UDP) packets. For the record, UDP connections lack a handshaking mechanism (unlike TCP), and therefore the IP address verification options are very limited. When this exploitation is in full swing, the volume of dummy packets exceeds the target server’s maximum capacity for processing and responding to requests.
This one is a variant of UDP Flood that specifically homes in on DNS servers. The malefactor generates a slew of fake DNS request packets resembling legitimate ones that appear to originate from a huge number of different IP addresses. DNS Flood is one of the hardest denial-of-service raids to prevent and recover from.
This is a common form of UDP Flood that targets a Voice over Internet Protocol (VoIP) server. The multitude of bogus VoIP requests sent from numerous IP addresses drain the victim server’s resources and knock it offline at the end of the day.
NTP Flood (NTP Amplification)
Network Time Protocol (NTP), one of the oldest networking protocols tasked with clock synchronization between electronic systems, is at the core of another DDoS attack vector. The idea is to harness publicly-accessible NTP servers to overload a target network with a large number of UDP packets.
Similarly to NTP, the Character Generator Protocol (CHARGEN) is an oldie whose emergence dates back to the 1980s. In spite of this, it is still being used on some connected devices such as printers and photocopiers. The attack comes down to sending tiny packets containing a victim server’s fabricated IP to devices with CHARGEN protocol enabled. In response, the Internet-facing devices submit UDP packets to the server, thus flooding it with redundant data.
Malefactors can exploit networked devices running Universal Plug and Play (UPnP) services by executing a Simple Service Discovery Protocol (SSDP) reflection-based DDoS attack. On a side note, SSDP is embedded in the UPnP protocol framework. The attacker sends small UDP packets with a spoofed IP address of a target server to multiple devices running UPnP. As a result, the server is flooded with requests from these devices to the point where it goes offline.
SNMP Flood (SNMP Amplification)
Tasked with harvesting and arranging data about connected devices, the Simple Network Management Protocol (SNMP) can become a pivot of another attack method. Cybercriminals bombard a target server, switch, or router with numerous small packets coming from a fabricated IP address. As more and more “listening” devices reply to that spoofed address, the network cannot cope with the immense quantity of these incoming responses.
When executing an HTTP Flood DDoS attack, an adversary sends ostensibly legitimate GET or POST requests to a server or web application, siphoning off most or all of its resources. This technique often involves botnets consisting of “zombie” computers previously contaminated with malware.
Recursive HTTP GET Flood
To perpetrate this attack, a malicious actor requests an array of web pages from a server, inspects the replies, and iteratively requests every website item to exhaust the server’s resources. The exploitation looks like a series of legitimate queries and can be difficult to identify.
Also referred to as Ping Flood, this incursion aims to inundate a server or other network device with numerous spoofed Internet Control Message Protocol (ICMP) echo requests or pings. Having received a certain number of ICMP pings, the network responds with the same number of reply packets. Since this capability to respond is finite, the network reaches its performance threshold and becomes unresponsive.
Misused Application Attack
Instead of using spoofed IP addresses, this attack parasitizes legitimate client computers running resource-intensive applications such as P2P tools. Crooks reroute the traffic from these clients to the victim server to bring it down due to excessive processing load. This DDoS technique is hard to prevent as the traffic originates on real machines previously compromised by the attackers.
IP Null Attack
This one is carried out by sending a slew of packets containing invalid IPv4 headers that are supposed to carry transport layer protocol details. The trick is that threat actors set this header value to null. Some servers cannot process these corrupt-looking packets properly and waste their resources trying to work out how to handle them.
This one involves a malware strain called Smurf to inundate a computer network with ICMP ping requests carrying a spoofed IP address of the target. The receiving devices are configured to reply to the IP in question, which may produce a flood of pings the server can’t process.
This DDoS technique follows a logic similar to the Smurf Attack, except that it deluges the intended victim with numerous UDP packets rather than ICMP echo requests.
Ping of Death Attack
To set this raid in motion, cybercrooks poison a victim network with unconventional ping packets whose size significantly exceeds the maximum allowed value (64 bytes). This inconsistency causes the computer system to allocate too many resources for reassembling the rogue packets. In the aftermath of this, the system may encounter a buffer overflow or even crash.
This attack stands out from the crowd because it requires very low bandwidth and can be fulfilled using just one computer. It works by initiating multiple concurrent connections to a web server and keeping them open for a long period of time. The attacker sends partial requests and complements them with HTTP headers once a while to make sure they don’t reach a completion stage. As a result, the server’s capability to maintain simultaneous connections is drained and it can no longer process connections from legitimate clients.
Low Orbit Ion Cannon (LOIC)
Originally designed as a network stress testing tool, LOIC can be weaponized in real-world DDoS attacks. Coded in C#, this open-source software deluges a server with a large number of packets (UPD, TCP, or HTTP) in an attempt to disrupt a target’s operation. This onslaught is usually backed by a botnet consisting of thousands of machines and coordinated by a single user.
High Orbit Ion Cannon (HOIC)
HOIC is a publicly accessible application that superseded the above-mentioned LOIC program and has a much bigger disruptive potential than its precursor. It can be used to submit a plethora of GET and HTTP POST requests to a server concurrently, which ends up knocking a target website offline. HOIC can affect up to 256 different domains at the same time.
ReDoS stands for “regular expression denial-of-service.” Its goal is to overburden a program’s regular expression implementation with instances of highly complex string search patterns. A malicious actor can trigger a regular expression processing scenario whose algorithmic complexity causes the target system to waste superfluous resources and slow down or crash.
This term denotes an attack that takes advantage of uncatalogued vulnerabilities in a web server or computer network. Unfortunately, such flaws are surfacing off and on, making the prevention a more challenging task.
Although distributed denial-of-service is an old school attack vector, it continues to be a serious threat to organizations. The sophistication and frequency of DDoS attacks are quickly increasing. According to recent estimates, the monthly number of such attacks exceeds 400,000. To top it off, cybercriminals keep adding new DDoS mechanisms to their repertoire and security providers aren’t always prepared to tackle them.
Another unnerving thing is that some techniques, including Low and High Orbit Ion Cannon, are open source and can be leveraged by wannabe criminals who lack tech skills. Such an attack may get out of hand and go way beyond the intended damage.
Recognizing DDoS attacks is critical to ensuring that your business and all data are protected. It is sometimes difficult to distinguish a slowdown in network performance or natural spikes in web traffic, so having the right people and technology in place should help you understand the difference.
To prevent DDoS attacks and minimize the impact, businesses should learn to proactively identify the red flags; have an appropriate response plan in place; make sure their security posture has no single point of failure, and continuously work on strengthening the network architecture.