CSE 363: Offensive Security - Denial of Service Attacks Study Notes

CSE 363: Offensive Security - Denial of Service Attacks

Overview

  • Instructor: Michalis Polychronakis

  • Institution: Stony Brook University

  • Date: 2026-02-26

Denial of Service (DoS) Attacks

  • Goal:

    • Harm the availability of service

    • Strain software, hardware, or network links beyond their capacity

    • Shut down or degrade service quality

    • Note: Not always a result of malicious attacks (e.g., flash crowds, "Slashdot effect", "hug of death")

  • Motives for DoS Attacks:

    • Protest/attention

    • Financial gain/damage

    • Revenge

    • Extortion

    • Evasion/diversion/cover

Financial Impact of DoS Attacks

  • Estimated Annual E-commerce Revenue:

    • Amazon: 115,879,000,000115,879,000,000

    • Revenue loss per hour during downtime: 13,219,12813,219,128

    • Walmart: 21,443,900,00021,443,900,000 (loss per hour: 2,446,2722,446,272)

    • Home Depot: 7,613,000,0007,613,000,000 (loss per hour: 868,464868,464)

    • Best Buy: 6,126,000,0006,126,000,000 (loss per hour: 698,832698,832)

    • Costco: 5,828,800,0005,828,800,000 (loss per hour: 664,920664,920)

    • Macy's: 6,108,500,0006,108,500,000 (loss per hour: 696,852696,852)

Characteristics of DoS Attacks

  • Attack Source:

    • Single source vs. Many sources

    • Distributed Denial of Service (DDoS): More than one source leading to an attack

  • Types of Impact:

    • Overload vs. Complete shutdown

    • Degradation of service vs. total disabling of software or equipment

  • Consequences of DoS Attacks:

    • Crash, restart, bricking, data loss, disk wipe, website defacement, etc.

  • Consumed Resources:

    • Network bandwidth, CPU, memory, sockets, buffer, disk storage, battery, human time

  • Amplification Factor:

    • Symmetric vs. Asymmetric attacks

    • Examples: Broadcast addresses, large protocol responses, exponential propagation, etc.

  • Algorithmic Complexity Attacks:

    • Induce worst-case behavior by triggering corner cases when processing input

  • Spoofing:

    • Hide the true sources of the attack

Lower Layer DoS Attacks

  • Physical Layer:

    • Wire cutting, equipment manipulation, physical destruction

    • RF jamming, signal interference

  • Link Layer:

    • MAC Flooding: Overload switch (CAM table exhaustion)

    • ARP Poisoning: Insert erroneous MAC-IP mappings into caches

    • DHCP Starvation: Consume all available addresses in the DHCP server

    • WiFi Deauthentication: Force disconnection from access points

Dynamic Host Configuration Protocol (DHCP)

  • Purpose:

    • Allow hosts to request configuration parameters (IP address, gateway, DNS server, etc.)

    • Operates using UDP with no authentication

  • DHCP Exhaustion:

    • Prevents clients from receiving IPs by consuming available addresses

    • Relies on MAC address spoofing

  • Rogue DHCP Server:

    • Provide incorrect information to clients

    • Can lead to Man-in-the-Middle (MitM) attacks

  • Defenses:

    • DHCP Snooping: Block bogus DHCP offers

    • Dynamic ARP Inspection (DAI): Prevent ARP spoofing through validation

Deauthentication Attacks

  • Mechanism:

    • Send spoofed deauth frames to access points

    • Disassociate clients from access points

    • Clients may connect to malicious ("evil twin") access points

  • Tools:

    • aireplay-ng, metasploit (deauth)

  • Auth Attacks:

    • Flood access point with spoofed random addresses to exhaust resources

Network Layer DoS

  • Flooding:

    • Bombard target with network packets

  • Types of Attacks:

    • Volumetric Attacks: Saturate available network bandwidth

    • Packet Rate Attacks: Overload packet processing engines

  • IP Spoofing:

    • Conceal attack source

    • Limited applications due to filtering not universally deployed

  • Broadcast Amplification:

    • One packet causes many responses

    • Example: ICMP Smurf Attack

Amplification Example: Smurf Attack

  • Mechanism:

    • Attacker sends spoofed ICMP Echo requests to victim's broadcast address

    • Victim is flooded with responses

  • Mitigation:

    • Configure hosts not to respond to ECHO requests

    • Configure routers to prevent forwarding packets to broadcast addresses

Transport Layer DoS

  • SYN Flooding:

    • Flood server with SYN requests

    • Causes resource exhaustion

    • Source IP may be spoofed

    • Involves connection termination via RST injection

  • Mitigation Techniques:

    • Drop old half-open connections

    • Implement SYN cookies

SYN Cookies

  • Concept:

    • Reply to SYN packets without maintaining per-connection state

    • Encode SYN queue entry state within a cookie

    • Restore state based on legitimate ACK packets

    • Default behavior in Ubuntu on SYN flood detection

TCP Connection Termination

  • Types of Terminations:

    • FIN: Indicates sender is done, but can still receive data

    • RST: Denotes both sending and receiving stops immediately

  • Impacts of RST Injection:

    • Can be used for censorship, blocking unwanted traffic

Application Layer DoS

  • Attacks on Application Layer:

    • Connection flooding and reflection

    • Exploit software vulnerabilities

    • Algorithmic complexity attacks to exploit worst-case scenarios

    • Spam as a form of resource attack

Telephony Denial of Service (TDoS)

  • Definition:

    • Overloading of communication networks with telephone calls

    • Can occur through both malicious intent and accidental events

    • Impacts emergency communication and response capabilities

Recent Trends in DDoS Attacks

  • Increase in DDoS attacks due to botnets

  • Use of compromised IoT devices for attack architecture (e.g., Mirai botnet)

  • Reports of record-breaking attack sizes (e.g., Cloudflare mitigated 5.6 Tbps attack)

Botnets and Their Role

  • Definition:

    • Networks of compromised systems controlled by an attacker

    • Can include PCs, mobile, and IoT devices

  • Challenges:

    • Difficult to combat due to scale and volume of attacks

    • Often rented through online markets

DDoS Traffic Growth

  • Key Drivers:

    • Extortion, theft, and gaming preferences for attack tools

DDoS Mitigation Strategies

  • General Defenses:

    • Packet filtering, capacity upgrading, etc.

    • Development of asymmetry in cost-benefit analysis of attacks vs defenses

  • Technical Solutions:

    • Ingress/egress filtering

    • Use of content delivery networks (CDNs) and redundancy

    • Explore overlay-based defensive systems

Conclusion

  • Understanding DoS and DDoS attack mechanisms is critical for cybersecurity.

  • Awareness and proactive mitigation strategies can limit the impact of such attacks.

  • Importance of continuing to develop and adapt defenses as the landscape evolves.