Fight an Army with an Army: Distributed Data Collection against

Fight an Army with an Army:
Distributed Data Collection against DDOS attacks
Ufuk Özkanlı & Mehmet Ercan Nergiz & Serdar Pehlivanoğlu
DDoS
Distributed denial-of-service
attacks have more than one
attack source that can be
organized
from
one
command center to
 Degrade
or
completely
disable a system
o by overloading critical
resources of the target
system
o or
by
exploiting
software bugs.
Problem Definition
Distributed Data Collection:
 Multiple Data Generator Nodes (G-Nodes) continously generates data
 Multiple Data Consumer Nodes (C-Nodes) fetch data
Application Areas
 Applications that collect data securely (Online voting, etc.).
 Making community supported data services (e.g., wikileaks, wikipedia,
stackoverflow, eksisozluk) resistant to DDOS attacks with the help of
volunteers.
 Accessing data in WEBsite when DNS addresses are blocked
Solution
 We introduce a static distributed network of volunteer S-Nodes that
acts as a buffer between G-Nodes and C-Nodes.
o S-Nodes collect data from G-Nodes and
o C-Nodes fetch data from S-Nodes
 Our topology satisfies k connectivity thus resistant to adversaries that
can attack at most k nodes.
Centralized
Problem Characteristic
 Attack Source
o Out of network attacks (Passive)
o insider attack (Active)
 Network Topology
o Static (No change of topology during the attack)
o Dynamic
 Deterministic
 Probabilistic
 Partially or Fully keeping data on S-Nodes
Capacity
1
Powerful
Trusted Server
DDoS
Resistancy
DDOS vulnerable
Cost
Less Expensive
Distributed
Volunteer Based
Distibution
100- ∞
Mobile Weak
S-Nodes
DDoS resistant to
DDoS resistant to
medium scale
Medium to Large
attacks
Scale attacks
Free
Expensive
(Volunteer based)
5-10 Powerful
Trusted Server