Inferring Internet Denial-of- Service Activity David Moore, Geoffrey M Voelker, Stefan Savage Presented by Yuemin Yu – CS290F – Winter 2005
Outline Motivation Attack types Backscatter analysis Results Conclusion
Motivation “How to prevalent are DOS attacks today on the internet?” Nature of the current treats Longer term analyses of trends and recurring patterns of attacks Publish quantitative data about attacks
Attack Types Logic attacks Exploit software vulnerabilities Software patches Flooding attacks Distributed DoS Spoof source IP address randomly Exhaust system resources
Backscatter Attacker uses randomly selected source IP address Victim reply to spoofed source IP Results in unsolicited response from victim to third party IP addresses
Backscatter
Backscatter Analysis m attack packets sent n distinct IP address monitored Expectation of observing an attack: R’ Actual rate of attack: R extrapolated attack rate
Analysis Assumptions Address uniformity Spoof at random Uniformly distributed Reliable delivery Attack and backscatter traffic delivered reliably Backscatter hypothesis Unsolicited packets observed represent backscatter
Attack classifications Flow-based Based on target IP address and protocol Fixed time frame (Within 5mins of most recent packet) Event-based Based on target IP address only Fixed time frame
Data collection /8 network 2^24 IP 1/256 of internet address space
Data collections Collect data extract following information TCP flags ICMP payload Address uniformity Port settings DNS information Routing information
Response/Used Protocols
Rate of attack
Victims by ports
Attack Duration Cumulative - Probability Cumulative probability density
Top level domain
Victims by Hostnames
Autonomous System
Repeated Attacks
Conclusion Observed 12,000 attacks against more than 5,000 distinct targets. Distributed over many different domains and ISP Small # long attacks with large % of attack volume An unexpected amount of attacks targeting home, foreign, specific ISP
Thanks Questions?