The Crossfire Attack
Download
Report
Transcript The Crossfire Attack
The Crossfire
Attack
MIN SUK KANG, SOO BUM LEE, VIRGIL D. GLIGOR
E C E D E PA R T M E N T A N D C Y L A B
CARNEGIE MELLON UNIVERSITY
2013 IEEE Symposium on Security and Privacy
Outline
INTRODUCTION
THE CROSSFIRE ATTACK
ATTACK PERSISTENCE AND COST
EXPERIMENT SETUP AND RESULTS
RELATED WORK
CONCLUSION
2
Outline
INTRODUCTION
THE CROSSFIRE ATTACK
ATTACK PERSISTENCE AND COST
EXPERIMENT SETUP AND RESULTS
RELATED WORK
CONCLUSION
3
INTRODUCTION – Old DDoS
Typical attack:
floods server with HTTP, UDP, SYN, ICMP…… packets
Persistence:
Maximum: 2.5 days
Average: 1.5days
Adversary’s Challenge:
DDoS Attacks are either Persistent or Scalable to N Servers
N traffic to 1 server => high-intensity traffic triggers network detection
Detection not triggered => low-intensity traffic is insufficient for N srevers
4
INTRODUCTION – Crossfire Attack
Link flooding by botnets cannot be easily countered
Spoofed IP addresses.
Can flood links without using unwanted traffic.
Launch an attack with low-intensity traffic flows that cross a
targeted link at roughly the same time and flood it.
5
INTRODUCTION – Crossfire Attack
A link-flooding attack that degrades/cuts off network
connections of scalable N-server area persistently.
Scalable N-Server areas
N = small(e.g., 1-1000 servers), medium(e.g., all servers in a US state),
large(e.g., the West Coast of the US)
Persistent:
Attack traffic is indistinguishable from legitimate
Low-rate, changing sets of flows
Attack is “ moving target ” for same N-server area
Changing target links before triggering alarms
6
INTRODUCTION – Definitions
7
INTRODUCTION – 1 link crossfire
Attack flows => Indistinguishable from legitimate
8
INTRODUCTION – 1 link crossfire
Attack flows => Alarms not triggered
link-failure detection
latency,
Interior Gateway
Protocol(IGP) routers (OSPF)
Default waiting time: 40sec, Failure detection: 217 sec
Exterior Gateway Protocol(EGP) routers(BGP)
Default waiting time: 180sec, Failure detection : 1,076 sec
9
Outline
INTRODUCTION
THE CROSSFIRE ATTACK
ATTACK PERSISTENCE AND COST
EXPERIMENT SETUP AND RESULTS
RELATED WORK
CONCLUSION
10
THE CROSSFIRE ATTACK
11
THE CROSSFIRE ATTACK
Public servers :
To construct an attack topology centered at target area
Decoy servers:
To create attack flow
12
ATTACK - Step 1 : Link Map Construction
( 72% )
(1) Traceroute ( B->S )
(2) Link-Persistence
13
ATTACK - Step 2 : Attack setup
DR: Degradation Ratio
(1) Flow-Density Computation
(2) Target-Link Selection
14
ATTACK - Step 3 : Bot Coordination
(1) Attack-Flow Assignment
(2) Target-Link Flooding
15
Outline
INTRODUCTION
THE CROSSFIRE ATTACK
ATTACK PERSISTENCE AND COST
EXPERIMENT SETUP AND RESULTS
RELATED WORK
CONCLUSION
16
ATTACK PERSISTENCE AND COST
Data-Plane-Only Attack : Indefinite Duration
Link failure detection
Traffic engineering
Proactive Attack Techniques : Rolling Attack
Maintaining the same target links
Changes bot and decoy servers
Maintaining the same target area
Changes target links
17
ATTACK PERSISTENCE AND COST
Attack bots available from Pay-per Install (PPI) markets [2011]
In experiments : 49% in US or UK, 37% in Europe, 14% rest of the world
10 target links : can be as low as 107,200 bots.
Cost approximately $9K
18
Outline
INTRODUCTION
THE CROSSFIRE ATTACK
ATTACK PERSISTENCE AND COST
EXPERIMENT SETUP AND RESULTS
RELATED WORK
CONCLUSION
19
EXPERIMENT SETUP AND RESULTS
Bots:
1,072 traceroute nodes 620 PlanetLab nodes, 452 LG(Looking Glass) servers
20
EXPERIMENT SETUP AND RESULTS
Decoy servers:
552 institutions (i.e., universities and colleges ) on both the East Coast
(10 states) and West Coast (7 states) of the US
2737 public web servers within Univ1 in Pennsylvania
7411 public web servers within Univ2 in Massachusetts
21
EXPERIMENT SETUP AND RESULTS
Target Areas:
22
EXPERIMENT SETUP AND RESULTS
23
EXPERIMENT SETUP AND RESULTS
Link map
Run a traceroute six times to diagnose link persistence
24
EXPERIMENT SETUP AND RESULTS
25
EXPERIMENT SETUP AND RESULTS
Average rate when flooding 10 Target Links against Pennsylvania
26
Outline
INTRODUCTION
THE CROSSFIRE ATTACK
ATTACK PERSISTENCE AND COST
EXPERIMENT SETUP AND RESULTS
RELATED WORK
CONCLUSION
27
The Coremelt Attack
28
“Spamhaus” Attack
29
RELATED WORK
30
Outline
INTRODUCTION
THE CROSSFIRE ATTACK
ATTACK PERSISTENCE AND COST
EXPERIMENT SETUP AND RESULTS
CONCLUSION
31
CONCLUSION
Attack Characteristics
Undetectability at the Target Area.
Indistinguishability of Flows in Routers
Persistence
Flexibility
New DDoS Attack: The Crossfire Attack
Scalable & Persistent
Internet-scale experiment
Feasibility of the attack
High impact with low cost
32
Q&A
33