EU FP6 LOBSTER European Infrastructure for accurate network - - PowerPoint PPT Presentation

eu fp6 lobster
SMART_READER_LITE
LIVE PREVIEW

EU FP6 LOBSTER European Infrastructure for accurate network - - PowerPoint PPT Presentation

EU FP6 LOBSTER European Infrastructure for accurate network monitoring An IST Project http://www.ist-lobster.org/ personal view on the future of ero-day Worm Containment Herbert Bos Vrije Universiteit Amsterdam herbertb _AT_ cs.vu.nl


slide-1
SLIDE 1

An IST Project http://www.ist-lobster.org/

1

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

EU FP6 LOBSTER

European Infrastructure for accurate network monitoring

personal view on the future of

ero-day Worm Containment

Herbert Bos Vrije Universiteit Amsterdam

herbertb _AT_ cs.vu.nl http://www.ist-lobster.org/

slide-2
SLIDE 2

An IST Project http://www.ist-lobster.org/

2

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

What is LOBSTER?

  • FP6 Specific Support Activity (SSA)
  • Duration: 09/2004 – 12/06
  • Partners

– FORTH – Vrije Universiteit Amsterdam – TNO ICT – CESNET – UNINETT – FORTHnet – ALCATEL – TERENA – Symantec?

slide-3
SLIDE 3

An IST Project http://www.ist-lobster.org/

3

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

What is LOBSTER?

  • European Infrastructure for accurate network monitoring
  • Allows one to perform pan-European monitoring

– across organisations

  • High-speed

– specialised network cards – also: common NICs

  • Why?

– traffic classification – security

  • worms
  • DDoS

– performance – billing – management

slide-4
SLIDE 4

An IST Project http://www.ist-lobster.org/

4

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Privacy

  • a

s h a r e d m

  • n

i t

  • r

i n g i n f r a s t r u c t u r e ?

  • w

h a t a b

  • u

t p r i v a c y ? !

slide-5
SLIDE 5

An IST Project http://www.ist-lobster.org/

5

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

What is LOBSTER?

  • Data owners control

– which users may access which data – very flexible

slide-6
SLIDE 6

An IST Project http://www.ist-lobster.org/

6

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Passive Monitoring and Security

  • Intrusion Detection

– Are any of my computers compromised? – Is there any attacker trying to intrude into my network?

  • Large-scale Attack Detection – Detection of Epidemics

– DoS Attack detection (e.g., detect sharp increases in TCP/SYN packets) – Zero-day worm detection

  • e.g., detect lots of identical packets, never seen before, from several sources to

several destinations

  • e.g., unusual no. of connections from a single port to unique destinations
  • e.g., detect worm characteristics

– such as NOP sleds: long sequences of executable code

  • Network Telescopes

– monitor unused IP addresses – observe victims of DoS attacks

  • “back-scatter” traffic

– observe infected hosts – port scans

slide-7
SLIDE 7

An IST Project http://www.ist-lobster.org/

7

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Zero-day worm containment

  • Why do we need it?

– detect something new is on the loose – worms spread too fast for human intervention

  • Different worms in different forms

– fast slow – polymorphic immutable – wide spread narrow spread – stealth plain – multi-vector uni-vector

  • Worm structure

exploit payload

slide-8
SLIDE 8

An IST Project http://www.ist-lobster.org/

8

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Two tasks

  • Spot the bad guys

– network-based

  • content-based: EarlyBird
  • flow-based: VirusThrottling

– host-based

  • honeypots
  • end-users (systrace)
  • Stop them!

– filters for networks

  • snort
  • VirusThrottle

– filters for hosts

  • Self-Certifying Alerts
  • can be fast (certainly flow-based)
  • protects many hosts
  • handles polymorphism
  • can be very accurate

(no false positives)

  • may handle polymorphism
  • handles polymorphism
  • protects many hosts
  • polymorphism
  • few false positives
  • some polymorphism?
slide-9
SLIDE 9

An IST Project http://www.ist-lobster.org/

9

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Two tasks

  • Spot the bad guys

– network-based

  • content-based: EarlyBird
  • flow-based: VirusThrottling

– host-based

  • honeypots
  • end-users (systrace)
  • Stop them!

– filters for networks

  • snort
  • VirusThrottle

– filters for hosts

  • Self-Certifying Alerts
  • false positives
  • what to do with encryption?
  • false positives
  • slow
  • needs a certain amount of luck
  • need real services for accuracy
  • false positives
  • encryption/polymorphism will kill us
  • false positives
  • pretty slow
  • can we rely on end-users?
slide-10
SLIDE 10

An IST Project http://www.ist-lobster.org/

10

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

My conclusion (1/4)

  • detection

– network-based

  • behaviour-based

– first indication

  • content-based:

– weed out known and old threats – first indication for new threats

– host-based

  • inaccurate behaviour based: first indication
  • accurate behaviour based:

– zero-day detection – verification

  • should not handle full streams
slide-11
SLIDE 11

An IST Project http://www.ist-lobster.org/

11

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

My conclusion (2/4)

  • blocking

– network-based

  • behaviour-based:

– no (unless exceptional circumstances)

  • content-based:

– weed out known and old threats – first indication for new threats

– host-based

  • good place for filtering, but scope of protection limited
  • end-host, so filtering should be fairly efficient
slide-12
SLIDE 12

An IST Project http://www.ist-lobster.org/

12

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

My conclusion (3/4)

  • future of network-based content inspection for

zero-day worm detection

slide-13
SLIDE 13

An IST Project http://www.ist-lobster.org/

13

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

My conclusion (4/4)

  • passive monitoring still needed, but role is

changing

– redirect traffic – sample traffic – first-pass detection – first-pass filtering – behaviour-based detection

  • explore

– multi-tier detection – multi-tier filtering – integrated approaches – cocktail-drugs for Internet diseases?

slide-14
SLIDE 14

An IST Project http://www.ist-lobster.org/

14

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Argos Emulator

Fingerprinting zero-day attacks and using advertised honeypots (or: guarding the heifer without falling asleep)

noah

slide-15
SLIDE 15

An IST Project http://www.ist-lobster.org/

15

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Argos Overview

  • Platform for next generation honeypots

– High-interaction, advertised, safe

  • Detection of most common vulnerabilities

– Control, code injection, function argument attacks

  • Emulate + protect entire PC systems

– OS agnostic, run on commodity hardware

  • Generate host and network intrusion prevention

signatures

– Protect even uncooperative users

  • Joint development with Dutch DeWorm project (VU)

noah

slide-16
SLIDE 16

An IST Project http://www.ist-lobster.org/

16

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Argos Overview

Argos emulator Guest OS Applications NIC Forensics

Detect attack and log state

Host OS

Correlate data Signature

Signature post-processing

Log

noah

slide-17
SLIDE 17

An IST Project http://www.ist-lobster.org/

17

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Argos Overview

Argos emulator Guest OS Applications NIC Forensics

Detect attack and log state

Host OS

Correlate data Signature

Signature post-processing

Log

noah

slide-18
SLIDE 18

An IST Project http://www.ist-lobster.org/

18

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Development up to Present

  • Based on the Qemu emulator
  • Track network data throughout execution
  • Detect illegal uses of network data

– Jump targets, function pointers, instructions, system call arguments

  • Forensics to generate signatures

– Export emulator state, inject “forensics” shellcode

noah

skip boring details

slide-19
SLIDE 19

An IST Project http://www.ist-lobster.org/

19

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Network Data Tracking

  • Tagging network data

as “tainted”

RAM Virtual NIC PORT I/O EBX ECX EDX EAX EBX

noah

slide-20
SLIDE 20

An IST Project http://www.ist-lobster.org/

20

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Network Data Tracking

  • Tagging network data

as “tainted”

  • Tracking “tainted” data

– ALU operations

RAM EBX ECX EDX EAX EBX ADD EAX, EBX EAX

noah

slide-21
SLIDE 21

An IST Project http://www.ist-lobster.org/

21

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Network Data Tracking

  • Tagging network data

as “tainted”

  • Tracking “tainted” data

– ALU operations

RAM EBX ECX EDX EAX EBX ADD EAX, EBX EAX XOR EBX, EBX

noah

slide-22
SLIDE 22

An IST Project http://www.ist-lobster.org/

22

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Network Data Tracking

  • Tagging network data

as “tainted”

  • Tracking “tainted” data

– ALU operations – MMU operations

RAM EBX ECX EDX EAX ADD EAX, EBX EAX XOR EBX, EBX ST A, EAX A

noah

slide-23
SLIDE 23

An IST Project http://www.ist-lobster.org/

23

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Identifying Attacks

  • Jump targets

RAM EBX ECX EDX EAX EAX A STACK JMP EAX ALERT

noah

slide-24
SLIDE 24

An IST Project http://www.ist-lobster.org/

24

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Identifying Attacks

  • Jump targets
  • Function calls

RAM EBX ECX EDX EAX EAX A STACK JMP EAX ALERT CALL EAX

noah

slide-25
SLIDE 25

An IST Project http://www.ist-lobster.org/

25

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Identifying Attacks

  • Jump targets
  • Function calls
  • Returns

RAM EBX ECX EDX EAX EAX A STACK JMP EAX ALERT CALL EAX RET

noah

slide-26
SLIDE 26

An IST Project http://www.ist-lobster.org/

26

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Identifying Attacks

  • Jump targets
  • Function calls
  • Returns
  • Code injection

RAM EBX ECX EDX EAX EAX A STACK JMP EAX ALERT CALL EAX RET JMP A

noah

slide-27
SLIDE 27

An IST Project http://www.ist-lobster.org/

27

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Identifying Attacks

  • Jump targets
  • Function calls
  • Returns
  • Code injection
  • System calls

RAM EBX ECX EDX EAX EAX A STACK JMP EAX ALERT CALL EAX RET JMP A INT 0x80

noah

slide-28
SLIDE 28

An IST Project http://www.ist-lobster.org/

28

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

Forensics

  • Emulator state (registers, “tainted” memory)
  • Injected shellcode data

– Process information (e.g. PID) – Extraction of probable target port PID Name Port

  • Network trace

noah

slide-29
SLIDE 29

An IST Project http://www.ist-lobster.org/

29

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

“tainted” memory data

Signature Generation

Network streams Alert value Origin of value in memory Forensics data Select stream Signature

noah

slide-30
SLIDE 30

An IST Project http://www.ist-lobster.org/

30

Herbert Bos, VU, http://www.cs.vu.nl/~herbertb

noah

http://www.few.vu.nl/argos