Chapter 8

Deworming

Deworming Deworming? Here, we discuss…1. Defenses2. Capture and containment3. Automatic counter-measures

Example Network

Example Network Perimeter computer

o Dedicated router or other computero Shuffles packets back and fortho May also perform defensive tasks

User --- self-explanatory Internal network

o Some critical machines to protect Honeypot is optional (discussed

later)

Defenses We consider defenses from

perspective of…o Userso Hosts (i.e., users’ computers)o Perimeter

Defenses: User User education

o Email worms User might detect suspicious

activityo E.g., slowdown in network responseo So, users might act as a type of IDS

Defenses: Host Most effective host-based defenses

are the most mundane1. Apply patches2. Limit available services3. Defend against likely attack

vectors4. Anti-virus software

Patching Most worms

exploit known problems

Patching follows curve something like this o Curve never

reaches 0

Patching Many machines remain vulnerable

long after patch is available Some machines never patched Strangely, patching rate does not

increase much when highly publicized attack occurs

Patching Why so slow?

o Nobody available to install patcho Don’t know about patcho Some don’t want to install patch early

(Why?) Modern OSs use automatic

updateso Is this a good idea?

Limit Available Services Why limit services?1. Worms often exploit technical

weaknesses in servers2. Technical weaknesses likely to be

uniformly distributed in server code

3. Other reasons? Which services to shut down?

Specific Countermeasures

1. Canaries, ASLR, and other buffer overflow prevention/detection

2. Anomaly detection These are of little/no use against

new vulnerabilities or attacks Might not catch worms that use

social engineering to spread

Anti-Virus Software AV does look for worms But 3 problems limit effectiveness1. Rapidly spreading worms may not

be detected using signatures2. Integrity check and emulation

might not worko Worm could be injected into “clean”

code

3. Worm might reside in memory only

Memory Scanning Search for malware in memory Once upon a time this was easy,

now complicated due to…o Memory protection, which keeps

process memory separateo Virtual memory, which is “extra”

memory on disk, not readily available for scanning

Aside: Hash check code in memory…

Memory Scanning How to scan?

o Challenging due to memory protection and virtual memory

Use API available for debugging?o Avoids memory protection, but slow if

scan virtual memoryo Ideally, only scan memory that

changeso But not all processes can be

debugged…o …unless AV runs as part of kernel

Memory Scanning What to do when malware found?

o Terminate infected process?o Terminate infected thread(s)?o Patch process as it runs?

Scanning cannot be continuouso Opens a window of vulnerability

Philosophically speaking, is memory scanning a good idea?

Perimeter Perimeter is 1st line of defense Perimeter computer might

include…o Firewall and/oro IDS

Discuss firewalls and IDS next

Firewalls

Firewalls

Firewall decides what to let in to internal network and/or what to let out

Access control for the network

InternetInternalnetworkFirewall

Firewall as Secretary A firewall is like a secretary To meet with an executive

o First contact the secretaryo Secretary decides if meeting is importanto So, secretary filters out many requests

You want to meet chair of CS department?o Secretary does some filtering

You want to meet the POTUS?o Secretary does lots of filtering

Firewall Terminology No standard firewall terminology Types of firewalls

o Packet filter works at network layero Stateful packet filter transport

layero Application proxy application layer

Other terms often usedo E.g., “deep packet inspection”

Packet Filter Operates at network layer Can filters based on…

o Source IP addresso Destination IP addresso Source Porto Destination Porto Flag bits (SYN, ACK, etc.)o Egress or ingress

application

transport

network

link

physical

Packet Filter Advantages?

o Speed Disadvantages?

o No concept of stateo Cannot see TCP connectionso Blind to application data

application

transport

network

link

physical

Packet Filter Configured via Access Control Lists

(ACLs)o Different meaning than at start of Chapter 8

Allow Inside Outside Any 80 HTTP

Allow Outside Inside 80 > 1023 HTTP

Deny All All All All All

Action

Source IP

Dest IP

Source

Port

Dest Port Protoco

l

Q: Intention? A: Restrict traffic to Web browsing

Any

ACK

All

FlagBits

TCP ACK Scan Attacker scans for open ports thru

firewallo Port scanning is first step in many attacks

Attacker sends packet with ACK bit set, without prior 3-way handshakeo Violates TCP/IP protocolo ACK packet pass thru packet filter firewallo Appears to be part of an ongoing connectiono RST sent by recipient of such packet

TCP ACK Scan

Attacker knows port 1209 open thru firewall

A stateful packet filter can prevent thiso Since scans not part of established

connections

PacketFilter

Trudy InternalNetwork

ACK dest port 1207

ACK dest port 1208

ACK dest port 1209

RST

Stateful Packet Filter

Adds state to packet filter Operates at transport layer Remembers TCP

connections, flag bits, etc. Can even remember UDP

packets (e.g., DNS requests)

application

transport

network

link

physical

Stateful Packet Filter Advantages?

o Can do everything a packet filter can do plus...

o Keep track of ongoing connections (e.g., prevents TCP ACK scan)

Disadvantages?o Cannot see application data

o Slower than packet filtering

application

transport

network

link

physical

Application Proxy A proxy is something

that acts on your behalf Application proxy looks at

incoming application data Verifies that data is safe

before letting it in

application

transport

network

link

physical

Application Proxy Advantages?

o Complete view of connections and applications data

o Filter bad data at application layer (viruses, Word macros)

Disadvantages?o Speed

application

transport

network

link

physical

Application Proxy Creates a new packet before sending it

thru to internal network Attacker must talk to proxy and

convince it to forward message Proxy has complete view of connection Prevents some scans stateful packet

filter cannot next slides

Firewalk Tool to scan for open ports thru firewall Attacker knows IP address of firewall

and IP address of one system inside firewallo Set TTL to 1 more than number of hops to

firewall, and set destination port to N

If firewall allows data on port N thru firewall, get time exceeded error message o Otherwise, no response

Firewalk and Proxy Firewall

This will not work thru an application proxy (why?)

The proxy creates a new packet, destroys old TTL

Dest port 12345, TTL=4

Dest port 12344, TTL=4

Dest port 12343, TTL=4

Time exceeded

Trudy

Packetfilter

RouterRouterRouter

Deep Packet Inspection Many buzzwords used for firewalls

o One example: deep packet inspection

What could this mean? Look into packets, but don’t really

“process” the packetso Like an application proxy, but faster

Firewalls and Defense in Depth

Typical network security architecture

Internet

Intranet withadditionaldefense

PacketFilter

ApplicationProxy

DMZ

FTP server

DNS server

Web server

Intrusion Detection Systems

Intrusion Prevention Want to keep bad guys out Intrusion prevention is a

traditional focus of computer securityo Authentication is to prevent intrusionso Firewalls a form of intrusion

preventiono Virus defenses aimed at intrusion

prevention Like locking the door on your car

Intrusion Detection In spite of intrusion prevention, bad

guys will sometime be successful Intrusion detection systems (IDS)

o Detect attacks in progress (or soon after)o Look for unusual or suspicious activity

IDS evolved from log file analysis IDS is currently a hot research topic How to respond when intrusion

detected?o We don’t deal with this topic here…

Intrusion Detection Systems

Who is likely intruder?o May be outsider who got thru firewallo May be evil insider

What do intruders do?o Launch well-known attackso Launch variations on well-known attackso Launch new/little-known attackso “Borrow” system resourceso Use compromised system to attack others.

etc.

IDS Intrusion detection approaches

o Signature-based IDSo Anomaly-based IDS

Intrusion detection architectureso Host-based IDSo Network-based IDS

Any IDS can be classified as aboveo In spite of marketing claims to the contrary!

Host-Based IDS Monitor activities on hosts for

o Known attackso Suspicious behavior

Designed to detect attacks such aso Buffer overflowo Escalation of privilege, …

Little or no view of network activities

Network-Based IDS Monitor activity on the network for…

o Known attackso Suspicious network activity

Designed to detect attacks such aso Denial of service (DoS)o Network probes (e.g., port scanning)o Malformed packets, etc.

Some overlap with firewall Little or no view of host-base attacks Can have both host and network IDS

Signature Detection Example

Failed login attempts may indicate password cracking attack

IDS could use the rule “N failed login attempts in M seconds” as a signatureo So, if N or more failed login attempts in M

seconds, IDS warns of attack Note that such a warning is specific

o Admin knows what attack is suspectedo Easy to verify attack (or false alarm)

Signature Detection

Suppose IDS warns whenever N or more failed logins in M secondso Set N and M so false alarms not common

o Can do this based on “normal” behavior

But, if Trudy knows signature threshold, she can try N − 1 logins every M seconds…

Then signature detection could slow down Trudy, but might not stop her

Signature Detection

Many techniques used to make signature detection more robust

Goal is to detect “almost” signatures For example, if “about” N login attempts

in “about” M secondso Warn of possible password cracking attempto What are reasonable values for “about”?o Can use statistical analysis, heuristics, etc.o Must not increase false alarm rate too much

Signature Detection Advantages of signature detection

o Simpleo Detect known attackso Know which attack at time of detectiono Efficient (if reasonable number of

signatures) Disadvantages of signature detection

o Signature files must be kept up to dateo Number of signatures may become largeo Can only detect known attackso Variation on known attack may not be

detected

Anomaly Detection Anomaly detection systems look for

unusual or abnormal behavior There are (at least) two challenges

o What is normal for this system?o How “far” from normal is abnormal?

No avoiding statistics here!o mean defines normalo variance gives distance from normal to

abnormal

How to Measure Normal?

How to measure normal?o Must measure during “representative”

behavioro Must not measure during an attack…o …or else attack will seem normal!o Normal is statistical meano Must also compute variance to have

any reasonable idea of abnormal

How to Measure Abnormal?

Abnormal is relative to some “normal”o Abnormal indicates possible attack

Statistical discrimination techniques include o Bayesian statisticso Linear discriminant analysis (LDA)o Quadratic discriminant analysis (QDA)o Neural nets, hidden Markov models (HMMs),

etc. Fancy modeling techniques also used

o Artificial intelligenceo Artificial immune system principleso Many, many, many others

Anomaly Detection (1) Spse we monitor use of three commands:

open, read, close Under normal use we observe Alice:

open, read, close, open, open, read, close, … Of the six possible ordered pairs, we see

four pairs are normal for Alice,(open,read), (read,close), (close,open), (open,open)

Can we use this to identify unusual activity?

Anomaly Detection (1)

We monitor use of the three commands open, read, close

If the ratio of abnormal to normal pairs is “too high”, warn of possible attack

Could improve this approach by o Also use expected frequency of each pairo Use more than two consecutive commandso Include more commands/behavior in the

modelo More sophisticated statistical discrimination

Anomaly Detection (2) Over time, Alice

has accessed file Fn at rate Hn

H0 H1 H2 H3

.10 .40 .40 .10

Is recent activity normal for Alice? We compute S = (H0A0)2+(H1A1)2+…+(H3A3)2

= .02o We consider S < 0.1 to be normal, so this is normal

How to account for use that varies over time?

Recently, “Alice” has accessed Fn at rate An

A0 A1 A2 A3

.10 .40 .30 .20

Anomaly Detection (2)

To allow “normal” to adapt to new use, we update averages: Hn = 0.2An + 0.8Hn

In this example, Hn are updated… H2=.2.3+.8.4=.38 and H3=.2.2+.8.1=.12

And we now have new long-term averages:

H0 H1 H2 H3

.10 .40 .38 .12

Anomaly Detection (2) The updated

long term average is

H0 H1 H2 H3

.10 .40 .38 .12

Is this normal use? Compute S = (H0A0)2+…+(H3A3)2 = .0488

o Since S = .0488 < 0.1 we consider this normal And we again update the long term

averages:

Hn = 0.2An + 0.8Hn

Suppose new observed rates…

A0 A1 A2 A3

.10 .30 .30 .30

Anomaly Detection (2) The starting

averages were:

H0 H1 H2 H3

.10 .40 .40 .10

Statistics slowly evolve to match behavior This reduces false alarms… …but also opens an avenue for attack

o Suppose Trudy always wants to access F3

o Can she convince IDS this is normal for Alice?

After 2 iterations, averages are:

H0 H1 H2 H3

.10 .38.364

.156

Anomaly Detection (2) To make this approach more robust,

must incorporate the variance Can also combine N stats Si as, say,

T = (S1 + S2 + S3 + … + SN) / N

to obtain a more complete view of “normal”

Similar (but more sophisticated) approach is used in an IDS known as NIDES

NIDES combines anomaly & signature IDS

Anomaly Detection Issues Systems constantly evolve and so must

IDSo Static system would place huge burden on SAs o But evolving IDS makes it possible for attacker

to (slowly) convince IDS that an attack is normal

o Attacker may win by simply “going slow” What does “abnormal” really mean?

o Indicates there may be an attacko Might not be any specific info about “attack”o How to respond to such vague information?o In contrast, signature detection is very specific

Anomaly Detection Advantages?

o Chance of detecting unknown attacks Disadvantages?

o Cannot use anomaly detection alone…o …must be used with signature detectiono Reliability is unclearo May be subject to attacko Anomaly detection indicates “something

unusual”, but lacks specific info on possible attack

Anomaly Detection: The Bottom Line

Anomaly-based IDS is active research topic Many people have high hopes for its

ultimate success Often cited as key future security

technology Hackers are not convinced

o Title of a talk at Defcon: “Why Anomaly-based IDS is an Attacker’s Best Friend”

Anomaly detection is difficult and tricky As hard as AI?

Capture and Containment Defense --- keep worms out Capture and contain --- keep

worms ino That is, prevent further spreado Might “sacrifice” some machines

Early capture is usefulo Honeypot

Containment measures?o Reverse firewalls and throttling

Honeypots Designed to look like a good target Originally aimed at human

attackers But now, for viruses, worms, etc. Used by researchers and AV

vendorso AV vendor sees new attackso Researchers study new attacks,

attack patterns, malware development cycle, etc.

Honeypots Properties of honeypot?

o Must be complete enough to look realo Impossible for worm to “break out”o Easy to restore after attack

How is honeypot built?o Emulation --- many of same

difficulties as in virus detection

Honeypots How do you know if it’s a worm in

the honeypot?o Perhaps honeypot does not attract any

legitimate traffico This won’t attract passive scanning

worm For random scanning worm, need

lots of honeypots For “smart” worm, honeypot must

make itself look attractive --- how?

Honeypots What does honeypot do with

worm?o Samples for analysiso Early warningo Measure of worm activity

Honeypot might act as a “tarpit”o Respond slowly --- slow down rapid

wormo No response --- could be devastating

to “hit list” worm

Reverse Firewalls Filter outgoing traffic

o Same principles as inbound firewalls Host-based reverse firewall

o Can limit activity based on application Worm might still succeed…

o Use shared network directorieso Subvert previously-checked codeo Use social engineeringo Take advantage of false positives

Throttling Limit rate of

connectionso For

outbound connections

o Throttle all applications

Throttling Can be refined

o Allow more connections to those recently connected to

o Take success into accounto “Safe” connections delayed by a littleo Lots of waiting connections is

suspicious Load balancing/fair share?

o Might adversely affect fast worms

Automatic Countermeasures

How to respond automatically?o I.e., no human intervention requiredo Why would we want to do this?

Two questions to consider… How to detect worm activity? What countermeasures to take?

How to Detect Worms? Automatic detection, that is Already considered honeypots,

throttling issues, IDS, etc. Others?

o Lots of traffic on one port with many different destination IPs (Why?)

o Few DNS queries relative to number of connection attempts (Why?)

Worm Detected? Automatic countermeasures?

o Isolate affected machineso Isolate critical networkso Shut down targeted serverso Automatically insert filtering rules into

firewallso Drop traffic to affected servers

Risks of automatic countermeasures?