Transcript Slide 1
Objectives
Get a non-product-specific perspective onto
security in IT
Demystify the commonly used terminology –
know your RC2 from AES
Bring together various aspects of security into
an integrated whole
Make some simple recommendations
Agenda
Defining Integrated Security (level 200)
Some Techniques for Securing IT (level 250)
Recommendations (level 200)
Printed/online slides include a section that covers
security risk analysis process – they are self-explanatory
(7 easy slides – please read at your leisure)
Warning: this is a fast and furious A-to-Z type of a session. Attend at your own risk.
Defining Security
Security
Definition (Cambridge Dictionary of English)
Ability to avoid being harmed by any risk, danger or
threat
…therefore, in practice, an impossible goal
What can we do then?
Be as secure as needed
Ability to avoid being harmed too much by
reasonably predictable risks, dangers or threats
(Rafal’s Definition)
Assets
What we are securing?
Data
Services (i.e. business etc. applications or their
individually accessible parts)
This session is not about securing:
People (sorry), cables, carpets, typewriters and
computers (?!)
Indeed: we (IT people) will secure the data on the
computer or services it offers and we will often
request that a PC should be locked up with an armed
guard but how this is done is not really our business
Sometimes known as physical security
Digital Security as Extension of
Physical Security of Key Assets
Strong Physical
Security of KA
Weak Physical
Security of KA
Strong Physical
Security of KA
Strong Digital
Security
Strong Digital
Security
Weak Digital
Security
Good Security
Everywhere
Insecure
Environment
Insecure
Environment
Aspects of Security
Confidentiality
◄ Your data/service provides no useful information to
unauthorised people
Integrity
◄ If anyone tampers with your asset it will be immediately
evident
Authenticity
◄ We can verify that asset is attributable to its authors or
caretakers
Non-repudiation
◄ The author or owner or caretaker of asset cannot deny that
they are associated with it
Identity
◄ We can verify who is the specific individual entity
associated with your asset
Additional Aspects of Data
and Service Security
Authorisation
◄ It is clear what actions are permitted with respect to your
asset
Loss
◄ Asset is irrecoverably lost (or the cost of recovery is too
high)
Denial of access (aka denial of service)
◄ Access to asset is temporarily impossible
“Static” cryptography is useful but not sufficient:
Backups etc. needed
Behaviour (pattern) of access analysis needed
Cryptography
Using really hard mathematics to implement most of the
security aspects mentioned earlier
“Static”
Cannot detect or prevent problems arising from a
pattern of behaviour
Relies of physical security of Key Assets (such as
master private keys etc.)
Strength changes with time, depending on the power of
computers and developments in cryptanalysis
Behaviour (Pattern)
Analysis
Prohibits reaching an asset if history of access is out-ofpattern, e.g.:
Password lock-out after N unsuccessful attempts
Blocking packets at a router if too many come from a given
source
Stopping a user from seeing more than N records in a
database per day
Time-out of an idle secure session
“Active”
Cannot prevent unauthorised use of asset – still need
crypto
Can prevent legitimate access – need easy and secure
“unlock” mechanisms
Strength varies with sophistication on known attacks
Integrated Security
Security should be Integrated Security:
Static + Active
Across
All Your Assets
Based On
Risk Assessment
st
1
Conclusion
As 100% security is impossible, you need to
decide what needs to be secured and how well
it needs to be secured
In other words, you need:
Asset list
Risk impact estimate for each asset
Some Techniques for
Securing IT
What is Really Secure?
Look for systems
From well-know parties
With published (not secret!) algorithms
That generate a lot of interest
That have been hacked for a few years
That have been analysed mathematically
Absolutely do not “improve” algorithms yourself
Employ someone to attempt a break-in
Behaviour (Pattern)
Analysis
Fairly new area (with exceptions)
In addition to knowing your assets, you need to
know your perimeter (edge)
Do you?
Active access inspection and pattern matching
are the main techniques
Many Perimeters
External – Network Edge
Between you and internet
etc.
DMZ – De-militarized Zone
Between network edge and
all protected resources
Only minimal protection
possible
Default Security Zone
The traditional LAN
High Security Zone
“Network inside network”
For key assets
Perimeter (Edge) of Isolation
Assets physically not
connected to networks
Useful for some key assets
(e.g. master keys)
Network Edge
High
Default
DMZ
Isolation
Tools for Behaviour
Analysis
Traditional: Firewalls and Proxies around the perimeters
(edges)
Stateful packet inspection
Traditional: Limiting number of accesses to Key Assets
Password lock-outs
Newer: Event Analysis and Active Agents
Rules can be programmed into some security servers
(ISA) or monitoring tools (MOM)
Neural networks are showing some promise for outof-pattern detection
Basic Crypto Terminology
Plaintext
The stuff you want to secure, typically readable by humans
(email) or computers (software, order)
Ciphertext
Unreadable, secure data that must be decrypted before it
can be used
Key
You must have it to encrypt or decrypt (or do both)
Cryptanalysis
Hacking it by using science
Complexity Theory
How hard is it and how long will it take to run a program
Symmetric Key
Cryptography
Plain-text input
“The quick
brown fox
jumps over
the lazy
dog”
Cipher-text
Plain-text output
“AxCv;5bmEseTfid3)
fGsmWe#4^,sdgfMwi
r3:dkJeTsY8R\s@!q3
%”
“The quick
brown fox
jumps over
the lazy
dog”
Encryption
Decryption
Same key
(shared secret)
Symmetric Pros and Cons
Weakness:
Must agree the key beforehand
Securely pass the key to the other party
Strength:
Simple and really very fast (order of 1000 to
10000 faster than asymmetric mechanisms)
Super-fast if done in hardware (DES,
Rijndael)
Hardware is more secure than software,
so DES makes it really hard to be done in
software, as a prevention
Public Key
Cryptography
Knowledge of the encryption key doesn’t give
you knowledge of the decryption key
Receiver of information generates a pair of keys
Publish the public key in a directory
Then anyone can send him messages that only
she can read
Public Key Encryption
Clear-text Input
“The quick
brown fox
jumps over
the lazy
dog”
Cipher-text
Clear-text Output
“Py75c%bn&*)9|fDe^
bDFaq#xzjFr@g5=&n
mdFg$5knvMd’rkveg
Ms”
“The quick
brown fox
jumps over
the lazy
dog”
Encryption
public
Recipient’s
public key
Decryption
Different keys
private
Recipient’s
private key
Public Key Pros and Cons
Weakness:
Extremely slow
Susceptible to “known ciphertext” attack
Strength
Solves problem of passing the key
Hybrid Encryption (Real World)
Launch key
for nuclear
missile
“RedHeat”
is...
Symmetric
encryption
(e.g. DES)
User’s
public key
(in certificate)
RandomlyGenerated
symmetric
“session” key
RNG
*#$fjda^j
u539!3t
t389E *&\@
5e%32\^kd
Symmetric key
encrypted asymmetrically
(e.g., RSA)
Digital
Envelope
As above, repeated
for other recipients
or recovery agents
Digital
Envelope
Other recipient’s or
agent’s public key
(in certificate)
in recovery policy
Hybrid Decryption
*#$fjda^j
u539!3t
t389E *&\@
5e%32\^kd
Launch key
for nuclear
missile
“RedHeat”
is...
Symmetric
decryption
(e.g. DES)
Symmetric
“session” key
Recipient’s
private key
Asymmetric
decryption of
“session” key (e.g. RSA)
Digital envelope
contains “session”
key encrypted
using recipient’s
public key
Digital
Envelope
Session key must be
decrypted using the
recipient’s private
key
Digital Signatures
Want to give plain text data to someone, and
allow them to verify the origin
Integrity, authenticity & non-repudiation
Much more on this in my PKI session SEC390 at
16:45 in room 6 today
DES, IDEA, RC2, RC5
Symmetric
DES (Data Encryption Standard) is the most popular
Keys very short: 56 bits
Brute-force attack took 3.5 hours on a machine costing
US$1m in 1993. Today it probably is done real-time.
Triple DES (3 DES) not much more secure but may thwart
NSA
Just say no, unless value of data is minimal
IDEA (International Data Encryption Standard)
Similar to DES, but “not” from NSA
128 bit keys
RC2 & RC5 (by R. Rivest)
RC2 is older and RC5 newer (1994) - similar to DES and
IDEA
Rijndael
Standard replacement for DES for US government, and,
probably for all of us as a result…
Winner of the AES (Advanced Encryption Standard)
competition run by NIST (National Institute of Standards
and Technology in US) in 1997-2000
Comes from Europe (Belgium) by Joan Daemen and
Vincent Rijmen. “X-files” stories less likely (unlike DES).
Symmetric block-cipher (128, 192 or 256 bits) with variable
keys (128, 192 or 256 bits, too)
Fast and a lot of good properties, such as good immunity from
timing and power (electric) analysis
Construction deceptively similar to DES (S-boxes, XORs etc.)
but really different
CAST and GOST
CAST
Canadians Carlisle Adams & Stafford Tavares
64 bit key and 64 bit of data
Chose your S-boxes
Seems resistant to differential & linear cryptanalysis and
only way to break is brute force (but key is a bit short!)
GOST
Soviet Union’s “version” of DES but with a clearer design
and many more repetitions of the process
256 bit key but really 610 bits of secret, so pretty much
“tank quality”
Backdoor? Who knows…
Careful with Streams!
Do NOT use a block cipher in a loop
Use a crypto-correct technique for treating
streams of data, such as CBC (Cipher Block
Chaining)
.NET Framework implements it as
ICryptoTransform on a crypto stream with
any supported algorithm
RC4
Symmetric
Fast, streaming encryption
R. Rivest in 1994
Originally secret, but “published” on sci.crypt
Related to “one-time pad”, theoretically most secure
But!
It relies on a really good random number generator
And that is the problem
RSA, DSA, ElGamal, ECC
Asymmetric
Very slow and computationally expensive – need a
computer
Very secure
Rivest, Shamir, Adleman – 1978
Popular and well researched
Strength in today’s inefficiency to factorise into prime
numbers
Some worries about key generation process in some
implementations
DSA (Digital Signature Algorithm) – NSA/NIST thing
Only for digital signing, not for encryption
Variant of Schnorr and ElGamal sig algorithm
ElGamal
Relies on complexity of discrete logarithms
ECC (Elliptic Curve Cryptography)
Really hard maths and topology
Better than RSA, in general and under a mass of research
Quantum Cryptography
Method for generating and passing a secret key or a random
stream
Not for passing the actual data, but that’s irrelevant
Polarisation of light (photons) can be detected only in a way
that destroys the “direction” (basis)
So if someone other than you observes it, you receive
nothing useful and you know you were bugged
Perfectly doable over 10-50km long fibre-optic link
But seems pretty perfect, if a bit tedious and slow
Don’t confuse it with quantum computing, which won’t be
with us for at least another 50 years or so, or maybe longer…
MD5, SHA
Hash functions – not encryption at all!
Goals:
Not reversible: can’t obtain the message from its hash
Hash much shorter than original
Two messages won’t have the same hash
MD5 (R. Rivest)
512 bits hashed into 128
Mathematical model still unknown
But it resisted major attacks
SHA (Secure Hash Algorithm)
US standard based on MD5
Diffie-Hellman, “SSL”,
Certs
Methods for key exchange
DH is very clever since you always generate a new “keypair” for each asymmetric session
STS, MTI, and certs make it even safer
Certs (certificates) are the most common way to
exchange public keys
Foundation of Public Key Infrastructure (PKI)
SSL uses a protocol to exchange keys safely
See session on PKI
Cryptanalysis
Brute force
Good for guessing passwords, and some 40-bit symmetric
keys (in some cases needed only 27 attempts)
Frequency analysis
For very simple methods only (US mobiles)
Linear cryptanalysis
For stronger DES-like, needs 243 plain-cipher pairs
Differential cryptanalysis
Weaker DES-like, needs from 214 pairs
Power and timing analysis
Fluctuations in response times or power usage by CPU
Breaking It on $10 Million
Symme-tric
Key
ECC Key
RSA Key
Time to
Break
Machines
Memory
56
112
420
< 5 mins
10000
Trivial
80
160
760
600
months
4300
4GB
96
192
1020
3 million
years
114
170GB
128
256
1620
10E16
years
0.16
120TB
From a report by Robert Silverman, RSA Laboratories, 2000
Some Recommendations
Strong Systems
It is always a mixture! Changes all the time…
Symmetric:
Min. 128 bits for RC2 & RC5, 3DES, IDEA, carefully
analysed RC4, 256 bit better
Asymmetric:
RSA, ElGamal, Diffie-Hellman (for keys) with
minimum 1024 bits (go for the maximum, typically
4096, if you can afford it)
Hash:
Either MD5 or SHA but with at least 128 bit results,
256 better
Weak Systems
Anything with 40-bits (including 128 and 56 bit versions
with the remainder “fixed”)
CLIPPER
A5 (GSM mobile phones outside US)
Vigenère (US mobile phones)
Dates from 1585!
Unverified certs with no trust
Weak certs (as in many “class 1” personal certs)
Summary
Decide what to secure and how
Have someone fulfil the role of CSO (Chief
Security Officer)
Combine static crypto-based security with
active behaviour (pattern) analysis
Use reasonably strong security mechanisms
Balance security against accessibility
Resources & Reading
Visit www.microsoft.com/security
Attend sessions on PKI (incl. SEC390)
For more detail, read:
Applied Cryptography, B. Schneier, John Wiley & Sons,
ISBN 0-471-12845-7
Foundations of Cryptography, O. Goldereich,
www.eccc.uni-trier.de/eccc-local/ECCCBooks/oded_book_readme.html
Handbook of Applied Cryptography, A.J. Menezes, CRC
Press,
ISBN 0-8493-8523-7
PKI, A. Nash et al., RSA Press, ISBN 0-07-213123-3
Cryptography in C and C++, M. Welschenbach, Apress,
ISBN 1-893115-95-X (includes code samples CD)
evaluations…
Please don’t
forget to
complete your
online Evaluation
Form
Risk Analysis for IT Security
A Bonus Section for Your Reading
Pleasure
Examples
Asset:
Internal mailbox of your Managing Director
Risk Impact Estimate (examples!)
Risk of loss: Medium impact
Risk of access by staff: High impact
Risk of access by press: Catastrophic impact
Risk of access by a competitor: High impact
Risk of temporary no access by MD: Low impact
Risk of change of content: Medium impact
Creating Your Asset List
List all of your named assets starting with the
most sensitive
Your list won’t ever be complete, keep updating
as time goes on
Create default “all other assets” entries
Divide them into logical groups based on
their probability of attacks or the risk of their
“location” between perimeters
Risk Impact Assessment
For each asset and risk attach a measure of impact
Monetary scale if possible (difficult) or relative numbers
with agreed meaning
E.g.: Trivial (1), Low (2), Medium (3), High (4),
Catastrophic (5)
Ex:
Asset: Internal MD mailbox
Risk: Access to content by press
Impact: Catastrophic (5)
Risk Probability
Assessment
Now for each entry measure probability the loss
may happen
Real probabilities (difficult) or a relative scale
(easier) such as: Low (0.3), Medium, (0.6), and
High (0.9)
Ex:
Asset: Internal MD mailbox
Risk: Access to content by press
Probability: Low (2)
Risk Exposure and Risk
List
Multiply probability by impact for each entry
Exposure = Probability x Impact
Sort by exposure
High-exposure risks need very strong security
measures
Lowest-exposure risks can be covered by default
mechanisms or ignored
Example:
Press may access MD mailbox:
Exposure = P(Low=0.3) x I(Catastrophic=5) = 1.5
By the way, minimum exposure is 0.3 and maximum is
4.5 is our examples
Mitigation and
Contingency
For high-exposure risks have a plan:
Mitigation: Reduce its probability or impact (so
exposure)
Transfer: Make someone else responsible for
the risk
Avoidance: avoid the risk by not having the
asset
Contingency: what to do if the risk becomes
reality
nd
2
Conclusion
Security risk management is an ongoing activity
which requires someone to be responsible for it
Who?
Your CSO – Chief Security Officer
Do you have one?