Security - Computer Science Department

advertisement
CMSC 414
Computer and Network Security
Jonathan Katz
Course Organization
Administrative
 Me
 TA
 Contact information, office hours, listed on course
webpage
Course webpage
http://www.cs.umd.edu/~jkatz/security/f09
 Syllabus
– Subject to change…
– Assigned readings and videos
• Will try to post by Friday for the following week
• Read in advance and come prepared to discuss
– Additional (optional) readings
 Homeworks distributed from the course webpage
 Check frequently for announcements
Class readings
 Material posted on the course webpage is fair
game for the exams, even if not covered in class
 Material covered in class is fair game for the
exams, even if not listed on the webpage
Textbook
 No required text
 Several good texts out there
– Will list on the course webpage
 Will supplement lectures with other readings
(distributed on class webpage)
Course requirements
 Homeworks
– 3-5 programming assignments
– Possibly 1-2 written assignments
 I expect students have access to a computer/laptop
capable of running a hypervisor
– VM player for Windows/linux (free download)
– VMware Fusion for MACs ($49.99 for academic
license), other free options may be available
– Occasional in-class exercises
Labs (tentative)
 Crypto
 Building a secure protocol
 Buffer overflow
 Web security
Piazza
http://piazza.com
 For your benefit
 Questions about lecture/readings
 Homework questions
 News items
– I encourage you to post links to news of interest!
Class participation
 Please!
Syllabus (tentative)
Syllabus I
 Introduction…
– A broad perspective on security
 Cryptography
– The basics (take CMSC 456 or read my book for more)
• If you took 456 with me, you can skip
– Cryptography is not the whole solution…
– …but it is an important part of the solution
– Along the way, we will see why cryptography can’t
solve all security problems
Syllabus II
 Network security I
– Identity, PKI
– Authentication and key exchange protocols
– Password and biometric authentication
– Anonymity and pseudonymity
– Privacy
Syllabus II
 System security
– General principles
– Security policies
– Access control
– OS security
– “Trusted computing”
 Programming language security
– Buffer overflows, input validation errors
– Viruses/worms
– Web security
Syllabus IV
 Privacy/anonymity
– Database security
– Anonymous communication
– Privacy in social networks
 Network security in the real world
– Some real-world protocols (IPSec/SSL)
– Security of network infrastructure (routing, DNS,
TCP/IP, DDos attacks, …)
Overview
Introduction and overview
 What is computer/network security? Why is it
important?
 Course philosophy and goals
 A broad perspective on “computer security”
Computer security is important…
 Several high-profile hacks in past years
– Number of vulnerabilities/attacks increasing
 Cyberwarfare
 Increasing gov’t and academic interest
 Just read the news…
Cybercrime
 e.g., botnets
 Washington Post, “Invasion of the Computer
Snatchers” (2006):
–
–
–
–
High-school dropout
Breaks into 2000 computers in 6 hours (while sleeping)
$6,800 per month; 2 minutes of work per day
$2B industry (annual)
Thoughts
 Why is the problem so difficult?
 What can be done about it?
“Security”
 Most of computer science is concerned with
achieving desired behavior
 Security is concerned with preventing undesired
behavior
– Different way of thinking!
– An enemy/opponent/hacker/adversary who is actively
and maliciously trying to circumvent any protective
measures you put in place
One illustration of the difference
 Software testing determines whether a given
program implements a desired functionality
– Test I/O characteristics
– Q/A
 How do you test whether a program does not
allow for undesired functionality?
– Penetration testing helps, but only up to a point
Why is computer security so hard?
 Computer networks are “systems of systems”
– Your system may be secure, then the environment changes
 Too many things dependent on a small number of systems
 Society is unwilling to trade off features for security
 Ease of attacks
–
–
–
–
Cheap
Distributed, automated
Anonymous
Insider threats
 Security not built in from the beginning
 Humans in the loop…
 Computers ubiquitous…
Computers are everywhere…
 …and can always be attacked
 Electronic banking, social networks, e-voting
 iPods, iPhones, PDAs, RFID transponders
 Automobiles
 Appliances, TVs
 (Implantable) medical devices
 Cameras, picture frames(!)
– See http://www.securityfocus.com/news/11499
A naïve view
password
In reality…
 Where does security end?
password
forgot password?
One good attack
 Use public records to figure out someone’s
password, or to get it from tech support
– E.g., hacked email account of Sarah Palin
 The password-recovery mechanism is part of the
system!
– The password-recovery mechanism may be the most
vulnerable point to attack
Computer security is not just about
computers
 What is “the system”?
 Physical security
 Social engineering
– Bribes for passwords
– Phishing
 “External” means of getting information
– Legal records, trash cans
 User education…
 Security is a process, not a product…(!)
Security is interdisciplinary
 Draws on all areas of CS
– Theory (especially cryptography)
– Networking
– Programming languages/compilers
– Operating systems
– Databases
– AI/learning theory
– Computer architecture / hardware
– HCI, psychology
Security mindset
 Learn to think with a “security mindset” in general
– What is “the system”?
– How could this system be attacked?
• What is the weakest point of attack?
– How could this system be defended?
• What threats am I trying to address?
• How effective will a given countermeasure be?
• What is the trade-off between security, cost, and usability?
An example: airline security
 Ask: what is the cost (economic and otherwise) of
current airline security?
 Ask: do existing rules (e.g., banning liquids) make
sense?
 Ask: are the tradeoffs worth it?
– (Why do we not apply the same rules to train travel?)
– (Would spending money elsewhere be more effective?)
 Ask: how would you get on a plane if you were on
the no-fly list?
– (I will not give you the answer – you can find it online)
– This is a thought experiment only!
Computer security is not just about
“security”
 Prevention…
 Detection, response, audit
– How do you know when you are being attacked?
– How quickly can you stop the attack?
– Attribution: can you identify the attacker(s)?
– Can you prevent the attack from recurring?
 Recovery
– Can be much more important than prevention
 Economics, insurance, risk management…
 Security is a process, not a product…
A naïve view
 Achieve “absolute” security
In reality…
 Absolute security is easy to achieve!
– How…?
 Absolute security is impossible to achieve!
– Why…?
 Good security is about risk management
Security as a trade-off
 The goal is not (usually) “to make the system as
secure as possible”…
 …but instead, “to make the system as secure as
possible within certain constraints” (cost,
usability, convenience)
– Military vs. personal networks
 Must understand the existing constraints
– E.g., passwords…
Cost-benefit analysis
 Important to evaluate what level of security is
necessary/appropriate
– Cost of mounting a particular attack vs. value of attack
to an adversary
– Cost of damages from an attack vs. cost of defending
against the attack
– Likelihood of a particular attack
 Sometimes the best security is to make sure you
are not the easiest target for an attacker…
“More” security not always better
 “No point in putting a higher post in the ground
when the enemy can go around it”
 Need to identify the weakest link
– Security of a system is only as good as the security at
its weakest point…
 Security is not a “magic bullet”
 Security is a process, not a product
Summary
 “The system” is not just a computer or a network
 Prevention is not the only goal
– Cost-benefit analysis
– Detection, response, recovery
 Nevertheless…in this course, we will focus on
computer security, and primarily on prevention
– If you want to be a security expert, you need to keep the
rest in mind
Philosophy of this course
 We are not going to be able to cover everything
– We are not going to be able to even mention everything
 Main goals
– A sampling of many different aspects of security
– The security “mindset”
– Become familiar with basic acronyms (RSA, SSL, PGP,
etc.), and “buzzwords” (phishing, …)
– Become an educated security consumer
– Try to keep it interesting with real-world examples and
“hacking” projects
Course goals
 You will not be a security expert after this class
(after this class, you should realize why it would
be dangerous to think you are)
 But you should have a better appreciation of the
threats, and how to address some of them
“Trusting trust”
“Trusting trust”
 Consider a compiler that embeds a trapdoor into
anything it compiles
 How to catch?
– Read source code? (What if replaced?)
– Re-compile compiler?
 What if the compiler embeds the trojan code
whenever it compiles a compiler?
– (That’s nasty…)
 Change compiler source S
compiler(S) {
if (match(S, "login-pattern")) {
compile (login-backdoor)
return
}
if (match(S, "compiler-pattern")) {
compile (compiler-backdoor)
return
}
.... /* compile as usual */
}
“Trusting trust”
 Whom do you trust?
 Does one really need to be this paranoid??
– Probably not
– Sometimes, yes
 Shows that security is complex…and essentially
impossible
 Comes back to risk/benefit trade-off
Assigned readings
 Thompson’s article
 “Inside the Twisted Mind of the Security
Professional”
 “We are All Security Customers”
 “Information Security and Externalities”
 Chapter 1 of “Security Engineering”
Download