Escolar Documentos
Profissional Documentos
Cultura Documentos
Dan Suciu
Joint work with Gerome Miklau
1
Data Security
Dorothy Denning, 1982:
• Data Security is the science and study of
methods of protecting data (...) from
unauthorized disclosure and modification
• Data Security = Confidentiality + Integrity
2
Data Security
• Distinct from systems and network security
– Assumes these are already secure
• Tools:
– Cryptography, information theory, statistics, …
• Applications:
– An enabling technology
3
Outline
• Traditional data security
• Two attacks
• Data security research today
• Conclusions
4
Traditional Data Security
• Security in SQL = Access control + Views
• Security in statistical databases = Theory
5
[Griffith&Wade'76, Fagin'78]
Access Control in SQL
GRANT privileges ON object TO users
[WITH GRANT OPTIONS]
privileges = SELECT | INSERT | DELETE | . . .
object = table | attribute
REVOKE privileges ON object FROM users
[CASCADE ]
6
Views in SQL
A SQL View = (almost) any SQL query
• Typically used as:
CREATE VIEW pmpStudents AS
SELECT * FROM Students WHERE…
GRANT SELECT ON pmpStudents TO DavidRispoli
7
Summary of SQL Security
Limitations:
• No row level access control
• Table creator owns the data: that’s unfair !
Access control = great success story of the DB community...
… or spectacular failure:
• Only 30% assign privileges to users/roles
– And then to protect entire tables, not columns
8
Summary (cont)
• Most policies in middleware: slow, error prone:
– SAP has 10**4 tables
– GTE over 10**5 attributes
– A brokerage house has 80,000 applications
– A US government entity thinks that it has 350K
• Today the database is not at the center of the
policy administration universe
9
[Rosenthal&Winslett’2004]
[Adam&Wortmann’89]
Security in Statistical DBs
Goal:
• Allow arbitrary aggregate SQL queries
• Hide confidential data
SELECT name
FROM Patient Not OK
SELECT count(*) WHERE age=42
FROM Patients OK
and sex=‘M’
WHERE age=42 and diagnostic=‘schizophrenia’
and sex=‘M’
and diagnostic=‘schizophrenia’
10
[Adam&Wortmann’89]
Security in Statistical DBs
What has been tried:
• Query restriction
– Querysize control, queryset overlap control, query monitoring
– None is practical
• Data perturbation
– Most popular: cell combination, cell suppression
– Other methods, for continuous attributes: may introduce bias
• Output perturbation
– For continuous attributes only
11
Summary on Security in
Statistical DB
• Original goal seems impossible to achieve
• Cell combination/suppression are popular,
but do not allow arbitrary queries
12
Outline
• Traditional data security
• Two attacks
• Data security research today
• Conclusions
13
[Chris Anley, Advanced SQL Injection In SQL]
SQL Injection
Your health insurance company lets you see the claims online:
********
Password:
Now search through the claims :
Search claims by: Dr. Lee
SELECT…FROM…WHERE doctor=‘Dr. Lee’ and patientID=‘fred’
14
SQL Injection
Now try this:
Search claims by: Dr. Lee’ OR patientID = ‘suciu’;
…..WHERE doctor=‘Dr. Lee’ OR patientID=‘suciu’; ’ and patientID=‘fred’
Better:
Search claims by: Dr. Lee’ OR 1 = 1;
15
SQL Injection
When you’re done, do this:
Search claims by: Dr. Lee’; DROP TABLE Patients;
16
SQL Injection
• The DBMS works perfectly. So why is
SQL injection possible so often ?
• Quick answer:
– Poor programming: use stored procedures !
• Deeper answer:
– Move policy implementation from apps to DB
17
Latanya Sweeney’s Finding
• In Massachusetts, the Group Insurance
Commission (GIC) is responsible for
purchasing health insurance for state
employees
• GIC has to publish the data:
GIC(zip, dob, sex, diagnosis, procedure, ...)
18
Latanya Sweeney’s Finding
• Sweeney paid $20 and bought the voter
registration list for Cambridge
Massachusetts:
GIC(zip, dob, sex, diagnosis, procedure, ...)
VOTER(name, party, ..., zip, dob, sex)
19
Latanya Sweeney’s Finding
zip, dob, sex
• William Weld (former governor) lives in
Cambridge, hence is in VOTER
• 6 people in VOTER share his dob
• only 3 of them were man (same sex)
• Weld was the only one in that zip
• Sweeney learned Weld’s medical records !
20
Latanya Sweeney’s Finding
• All systems worked as specified, yet an
important data has leaked
• How do we protect against that ?
Some of today’s research in data security address breaches
that happen even if all systems work correctly
21
Summary on Attacks
SQL injection:
• A correctness problem:
– Security policy implemented poorly in the application
Sweeney’s finding:
• Beyond correctness:
– Leakage occurred when all systems work as specified
22
Outline
• Traditional data security
• Two attacks
• Data security research today
• Conclusions
23
Research Topics in Data Security
Rest of the talk:
• Information Leakage
• Privacy
• Finegrained access control
• Data encryption
• Secure shared computation
24
[Samarati&Sweeney’98, Meyerson&Williams’04]
Information Leakage:
kAnonymity
Definition: each tuple is equal to at least k1 others
Anonymizing: through suppression and generalization
First Last Age Race
Harry
* Stone 3050
34 AfrAm
John Reyser
R* 2040
36 Cauc
*
Beatrice
* Stone 3050
47 Afram
John Ramos
R* 2040
22 Hisp
*
Hard: NPcomplete for supression only
Approximations exists
25
[Miklau&S’04, Miklau&Dalvi&S’05,Yang&Li’04]
Information Leakage:
Queryview Security
Have data: TABLE Employee(name, dept, phone)
• The applications:
– many years away
27
Privacy
• “Is the right of individuals to determine for
themselves when, how and to what extent
information about them is communicated to
others” [Agrawal’03]
• More complex than confidentiality
28
Privacy
Involves: Example: Alice gives her email
to a web service
• Data
• Owner
• Requester alice@a.b.com
• Purpose
• Consent
Privacy policy: P3P
29
Hippocratic Databases
DB support for implementing privacy policies.
• Purpose specification
• Consent Hippocratic DB
• Limited use alice@a.b.com
• Limited retention
• …
Privacy policy: P3P
Protection against:
Sloppy organizations
30
× Malicious organizations [Agrawal’03, LeFevrey’04]
Privacy for Paranoids
• Idea: rely on trusted agents
alice@a.b.com aly1@agenthost.com
Agent
lice27@agenthost.com
Protection against:
Sloppy organizations foreign keys ?
Malicious attackers
31
[Aggarwal’04]
Summary on Privacy
• Major concern in industry
– Legislation
– Consumer demand
• Challenge:
– How to enforce an organization’s stated
policies
32
Finegrained Access Control
Control access at the tuple level.
• Policy specification languages
• Implementation
33
Policy Specification Language
No standard, but usually based on parameterized views.
CREATE AUTHORIZATION VIEW PatientsForDoctors AS
SELECT Patient.*
FROM Patient, Doctor
WHERE Patient.doctorID = Doctor.ID
and Doctor.login = %currentUser
Context
parameters
34
Implementation
SELECT Patient.name, Patient.age
FROM Patient
WHERE Patient.disease = ‘flu’
SELECT Patient.name, Patient.age
FROM Patient, Doctor
WHERE Patient.disease = ‘flu’
and Patient.doctorID = Doctor.ID e.g. Oracle
and Patient.login = %currentUser
35
Two Semantics
• The Truman Model = filter semantics
– transform reality
– ACCEPT all queries SELECT count(*)
– REWRITE queries FROM Patients
– Sometimes misleading results WHERE disease=‘flu’
• The nonTruman model = deny semantics
– reject queries
– ACCEPT or REJECT queries
– Execute query UNCHANGED
– May define multiple security views for a user
36
[Rizvi’04]
Summary of Fine Grained Access
Control
• Trend in industry: labelbased security
• Killer app: application hosting
– Independent franchises share a single table at
headquarters (e.g., Holiday Inn)
– Application runs under requester’s label, cannot
see other labels
– Headquarters runs Read queries over them
• Oracle’s Virtual Private Database
37
[Rosenthal&Winslett’2004]
Data Encryption for Publishing
Scientist wants to publish
medical research data on the Web
All authorized users: Kuser
Patient: Kpat
• Users and their keys: Doctor: Kdr
Nurse: Knu
Administrator : Kadmin
Doctor researchers may access trials
Nurses may access diagnostic
• Complex Policies: Etc…
What is the encryption granularity ?
38
[Miklau&S.’03]
Data Encryption for Publishing
Doctor: Kuser, Kdr
An XML tree protection:
Nurse: Kuser, Knu
Nurse+admin: Kuser, Knu, Kadm
<patient> Kuser
• Research:
– Hard theoretical security analysis
[Abadi&Warinschi’05]
40
Secure Shared Processing
• Alice has a database DBA
• Bob has a database DBB
• How can they compute Q(DBA, DBB), without
revealing their data ?
• Long history in cryptography
• Some database queries are easier than general case
41
[Agrawal’03]
Secure Shared Processing
Alice Bob
Task: find intersection
without revealing the rest
a b c d c d e
Compute oneway hash
h(c) h(d) h(e) h(a) h(b) h(c) h(d)
What’s wrong ?
42
[Agrawal’03]
Secure Shared Processing
Alice commutative encryption: Bob
h(x) = EA(EB(x)) = EB(EA(x))
a b c d c d e
EA EB
EA(a) EA(b) EA(c) EA(d) EB(c) EB(d) EB(e)
EB(c) EB(d) EB(e) EA(a) EA(b) EA(c) EA(d)
EA EB
h(c) h(d) h(e) h(a) h(b) h(c) h(d)
h(a) h(b) h(c) h(d)
h(c) h(d) h(e)
43
Summary on Secure Shared
Processing
• Secure intersection, joins, data mining
• But are there other examples ?
44
Outline
• Traditional data security
• Two attacks
• Data security research today
• Conclusions
45
Conclusions
• Traditional data security confined to one server
– Security in SQL
– Security in statistical databases
• Attacks possible due to:
– Poor implementation of security policies: SQL
injection
– Unintended information leakage in published data
46
Conclusions
• State of the industry:
– Data security policies: scattered throughout applications
– Database no longer center of the security universe
– Needed: automatic means to translate complex policies into
physical implementations
• State of research: data security in global data sharing
– Information leakage, privacy, secure computations, etc.
– Database research community has an increased appetite for
cryptographic techniques
47
Questions ?
48