Transcript CMSC 414 Computer (and Network) Security

CMSC 414
Computer and Network Security
Lecture 28
Jonathan Katz
Administrivia
 Final exam reminder + study guide
– DSS students contact me
– A-G in 3258 AVW; H-Z in classroom
 Course evaluations
– www.CourseEvalUM.umd.edu
SSL in wireshark
IPsec
Overview
 IPsec can provide security between any two
network-layer entities
– host-host, host-router, router-router
 Used widely to establish VPNs
 IPsec encrypts and/or authenticates network-layer
traffic, and encapsulates it within a standard IP
packet for routing over the Internet
Overview
 IPsec is a complex, over-engineered protocol
– Lots of un-needed features
 Interoperability is challenging
– Defeats the point of having a standard
 IPsec is less used than it should(?) be
Overview
 IPsec consists of two components
– IKE --- Used to establish a key
– AH/ESP --- Used to send data once a key is established
(whether using IKE or out-of-band)
 AH (authentication header)
– Data integrity, but no confidentiality
 ESP (encapsulating security payload)
– Data integrity + confidentiality
– (Other differences as well)
Security policy database
 Nodes maintain a table specifying what is required
for each incoming packet
– Drop
– Forward/accept without IPsec protection
– Require IPsec protection
• Auth only
• Enc only
• Both
 Decisions can be based on any information
contained in the packet
Security associations (SAs)
 When a node receives a packet, it needs to know
who it is from
– May be receiving IPsec traffic from multiple senders at
the same time -- possibly even with the same IP address
 An SA defines a network-layer unidirectional
logical connection
– For bidirectional communication, need two SAs
 The IPsec header indicates which security
association to use
Security associations (SAs)
 An SA contains crypto keys, the identity/IP
address of the other party, a sequence number, and
crypto parameters (algorithms, auth/enc/both)
IPsec: IKE
Overview of IKE
 IKE provides mutual authentication, establishes a
shared key, and creates an SA
 Assumes a long-term shared key, and uses this to
establish a session key (as well as to provide
authentication)
 Supported key types
– Public signature keys
– Public encryption keys
– Symmetric keys
IKE phases
 Phase 1: long-term keys used to derive a session
key (and provide authentication)
– Roughly analogous to SSL session
 Phase 2: the session key is used to derive SAs
– Roughly analogous to SSL connection
 In theory, can run phase 1 once, followed by
multiple executions of phase 2
• E.g., different flows between same endpoints
• Why not use same key for each?
– In practice, this rarely happens
Phase 1 session keys
 Two session keys are defined in phase 1
– One each for encryption/authentication
 These keys are used to protect the final phase 1
messages as well as all phase 2 messages
Key types
 As mentioned earlier…
 Why are there two PK options?
– Signature-based option
• Efficiency (can start protocol knowing only your own public
key, then get other side’s key from their certificate)
• Legal reasons/export control
– Encryption-based option
• Can be used to provide anonymity in both directions
 Adds tremendously to the complexity of
implementation
IKE phase 1
 Aggressive mode
– 3 messages
 Main mode
– 6 messages
– Additional features:
• Anonymity
• Negotiation of crypto parameters
Anonymity
 Protocols can be designed so that identities of the
parties are hidden from eavesdroppers
– Even while providing authentication!
 Can also protect anonymity of one side against
active attacks
– Whom to protect?
• Initiator: since responder’s identity is generally known…
• Responder: since otherwise it is easy to get anyone’s identity
Aggressive mode
 Alice sends ga, “Alice”, crypto algorithms
– Note that choices are restricted by this message
 Bob sends gb, choice of crypto algorithm, “proof”
that he is really Bob
– If Bob does not support any of the suggested
algorithms, he simply does not reply
– Note that there is no way to authenticate a refusal, since
no session key yet established
 Alice sends “proof” that she is Alice
 Derive shared key from gab
Main mode
 Negotiate crypto algorithms (2 rounds)
 Alice and Bob do regular Diffie-Hellman key
exchange (2 rounds)
 Alice sends encryption of “Alice” plus a proof that
she is Alice, using long-term secret keys plus
(keys derived from) gab
 Bob does similarly…
Crypto parameters…
 Choice of:
– Encryption method (DES, 3DES, …)
– Hash function (MD5, SHA-1, …)
– Authentication method (e.g., key type, etc.)
– Diffie-Hellman group (e.g., (g, p), etc.)
 A complete set of protocols (a security suite) must
be specified
Negotiating parameters
 Many protocols allow parties to negotiate
cryptographic algorithms and parameters
– Allows users to migrate to stronger crypto; increases
inter-operability (somewhat)
 But, opens up a potential attack if not
authenticated somehow…
 Also makes for more complicated
implementations
“Proofs of identity”
 Depend on which type of long-term shared key is
being used
 Similar (in spirit) to the authentication protocols
discussed in class
IPsec: AH/ESP
AH vs. ESP
 Two header types…
 Authentication header (AH)
– Provides integrity only
 Encapsulating security payload (ESP)
– Provides encryption + integrity
 Both provide cryptographic protection of
everything beyond the IP headers
– AH additionally provides integrity protection of some
fields of the IP header
Transport vs. tunnel mode
 Transport mode: add IPsec information between IP
header and rest of packet
– IP header | IPsec | [ packet ]
protected
 Designed for end-to-end secure communication
Firewalls and transport mode
 Transport mode may cause problems if there are
firewalls between the communicating hosts
– Firewalls can’t inspect higher-layer information, like
ports or applications
 Tunnel mode was suggested to overcome this…
– Traffic secured between the firewalls (or between one
firewall and the other end host)
– End host(s) can be oblivious to what is being done
Transport vs. tunnel mode
 Tunnel mode: keep original IP packet intact but
protect it; add new header information outside
– New IP header | IPsec | [ old IP header | packet ]
protected
– Can be used when IPSec is applied at intermediate
points along path (e.g., for firewall-to-firewall traffic)
• Treat the link as a secure tunnel
– New IP header different from old header since, e.g.,
src/dest have changed
Transport vs. tunnel mode
 Note that tunnel mode subsumes transport mode…
– …but transport mode is more efficient
 Tunnel mode also hides (some information about)
the communicating parties
More on AH
 AH provides integrity protection on header
– But some fields change en route!
 Immutable fields included in the integrity check
 Mutable but predictable fields are also included in
the integrity check
– The final value of the field is used
More on ESP
 ESP provides both confidentiality and integrity
– On data only, not header
• Header can’t be encrypted
More on AH vs. ESP
 ESP can already provide encryption and/or
authentication
 So why do we need AH?
– AH also protects the IP header
– Export restrictions
– Firewalls need some high-level data to be unencrypted
 None of these are compelling…
Intrusion detection
Prevention vs. detection
 Firewalls (and other security mechanisms) aim to
prevent intrusion
 IDS aims to detect intrusion in case it occurs
 Use both in tandem!
– Defense in depth, full prevention impossible
– Outsider vs. insider attacks
– The sooner intrusion is detected, the less the damage
– IDS can also be a deterrent, and can be use to detect
weaknesses in other security mechanisms
IDS tradeoff
 IDS based on the assumption that attacker
behavior is (sufficiently) different from legitimate
user behavior
 In reality, there will be overlap
– Some legitimate behavior may appear malicious
– Intruder can attempt to disguise their behavior as that of
an honest user
False positives/negatives
 False positive
– Alarm triggered by acceptable behavior
 False negative
– No alarm triggered by illegal behavior
 Always a tradeoff between the false positive and
false negative rate
False alarms?
 Say we have an IDS that is 99% accurate
– I.e., Pr[alarm | attack] = 0.99 and
Pr[no alarm | no attack] = 0.99
 An alarm goes off -- what is the probability that an
attack is taking place?
 To increase this probability, what should we focus
on improving??
False alarms
 Say the probability of an attack is 1/1000
 Use Bayes’ law:
Pr[attack | alarm]
= Pr[alarm | attack] Pr[attack] / Pr[alarm]
= 0.99 * 0.001 / (0.99 * 0.001 + 0.01 * 0.999)
≈ 0.001/(0.001 + 0.01) ≈ 0.1
 I.e., when an alarm goes off, 90% of the time it
will be a false alarm!
 How best to lower this number?
False alarms
 Improving Pr[alarm | attack] to 100% gives
Pr[attack | alarm] ≈ 0.1 (essentially unchanged)
 Improving Pr[no alarm | no attack] to 99.9% gives
Pr[attack | alarm] ≈ 0.5
Two types of IDS
 Signature-based ≈ looks for improper behavior
– Roughly analogous to blacklisting
 Anomaly-based ≈ looks for atypical behavior
– Roughly analogous to whitelisting
Signature (rule-based) detection
 Define a set of “bad patterns” (e.g., known exploit
characteristics, known bad events)
 Detect these patterns if they occur
Example rules
 Incoming packets with a certain pattern match a
known exploit
 Users should not read files in other users’
directories
 Users should not be logged in simultaneously
from more than one location
 Users do not make copies of system programs
 No incoming requests except to port 80
Anomaly detection
 Monitor behavior and compare to some “baseline”
behavior using statistical tests
– Look for deviations from “normal behavior”
 “Normal behavior” can be defined on a global
level or a per-user level
 “Normal behavior” can be specified by a human,
or learned automatically over time
Probability
density
function
Profile of Intruder
behavior
Profile of
authorized user
behavior
Overlap in observed or
expected behavior
Average
behaviour of
intruder
Average
behaviour of
authorized user
Measurable
behaviour
parameter
Metric
Model
Justification
Login frequency by date Mean and standard
and time
deviation
Intruders are more likely
to login during off-hours
Frequency of login at
different locations
Mean and standard
deviation
Intruders may login from
a location that a legitimate
user does not
Time since last login
Markov (time series)
Break-in to unused
account
Length of session
Mean and standard
deviation
Masquerader may run a
much shorter or longer
session
Large amount of data
copied to some location
Mean and standard
deviation
Detect attempt to copy
large amounts of sensitive
data
Password failures at
login
Unusual event/
operational
Detect attempt to guess
passwords
Two places IDS can be run
 Host-based IDS (HIDS)
 Network-based IDS (NIDS)
Host-based IDS
 Monitors events on a single host
 Can (potentially) observe the effects of an attack
(in addition to possibly detecting the attack itself)
 Can detect both internal and external intrusions
Distributed host-based IDS
 Combine information collected at many different
hosts in the network
 One or more machines in the network will collect
and analyze the network data
– Can correlate information across multiple hosts
– E.g., same event occurring simultaneously across all
machines might be suspicious
– Or, an event occurring on only one machine might be
suspicious
Network-based IDS
 Monitors traffic at selected points on the network
– Real time; packet-by-packet
 By looking at all network traffic, can potentially
get a global view
Sensor types
 Inline sensor
– Inserted in network path; all traffic passes through the
sensor
 Passive sensor
– Monitors a copy of network traffic
 Passive sensor more efficient; inline sensor can
block attacks immediately
Sensor placement
 Inside firewall?
– Can detect attacks that penetrate firewall
– Can detect firewall misconfiguration
– Can examine outgoing traffic more easily to detect
insider attacks
– Can configure based on network resources being
accessed (e.g., configure differently for traffic directed
to web server)
 Outside firewall?
– Can document attacks (types/locations/number) even if
prevented by firewall (can then be handled out-of-band)
Drawbacks of NIDS
 Cannot analyze encrypted traffic
 Cannot observe attack effects
Honeypots
 Decoy systems to lure potential attackers
– Divert attackers from critical systems
– Collect information about attacker’s activity
– Delay attacker long enough to respond
 Since honeypot is not legitimate, any access to the
honeypot is suspicious
 Can have honeypot computers, or even honeypot
networks
Honeypot placement
 Outside firewall
– Can detect attempted connections to unused IP
addresses, port scanning
– No risk of compromised system behind firewall
– Does not divert internal attackers
 Fully internal honeypot
– Catches internal attacks
– Can detect firewall misconfigurations/vulnerabilities
– If compromised, run the risk of a compromised system
Course summary
What should you take away from
this course (after the final)?
 Security mind-set
– Not limited to computers/networks!
 Security is complex
– Draws on many different disciplines
– Need to know what you are doing
 Security is hard, still evolving
– We did not cover some of the most important presentday attacks: spam, phishing, DDos, viruses, …
 Security is challenging…but fun!
Thank you!