Database Security Control Measures

We follow Elmasrı R., Navathe S.B., Fundamentals of Database

Systems, 5th Ed., Pearson, 2007, Chapter 23
There are four main control measures that are used to provide security of
data in databases:
 Access control
 Inference control
 Flow control
 Data encryption
Access control we have already considered and we shall extend it. Inference
control measures are used to prevent extraction (inference) of private
information from publicly available statistical databases such as salary
summary in universities’ departments. Flow control prevents information
from flowing in such a way that it reaches unauthorized users. Channels that
are pathways for information to flow implicitly in ways that violate the
security policy of an organization are called covert channels. A final control
measure is data encryption. We have also touched this issue.

Database Security and the DBA

The database administrator (DBA) is the central authority for managing a
database system. The DBA’s responsibilities include account creation
(access control as a whole), granting/revoking privileges to/from users
(discretionary authorization control) and classifying users and data in
accordance with the policy of the organization (mandatory authorization
control). The DBA has a DBA account, sometimes called a system or super-
user account, which provides powerful capabilities that are not made
available to regular database accounts and users.

Access Protection, User Accounts, and Database Audits

Each valid user gets from DBA an account number and password, which are
used for user authentication when he logs in the system. For each login
session, the DBMS can record the user’s account number and associate it
with the terminal from which the user logged in. All operations applied from
the terminal are attributed to the user’s account until the user logs off. It is
particularly important to keep track of update operations that are applied to
the database so that, if the database is tampered with, the DBA can
understand which user did the tampering. If any tampering of the database is
suspected, a database audit is performed, which consists of reviewing the

1
Access Protection, User Accounts, and Database Audits (Cont 1)

log file (audit file) to examine all accesses and operations applied to the
database during a certain time period.

Discretionary Access Control Based on Granting and Revoking

Privileges
There are two levels for assigning privileges to use the database system:
 The account level. At this level, the DBA specifies the particular
privileges that each account holds independently of the relations in the
database (e.g., CREATE SCHEMA and CREATE TABLE privileges)
 The relation (table) level. At this level, the DBA can control the
privilege to access each individual relation or view in the database
(user access control matrix, where the rows represent subjects (users,
accounts, programs) and the columns represent objects (relations,
records, columns, views, operations)
To control the granting and revoking of relation privileges, each relation R in
a database is assigned an owner account, which is typically the account that
was used when the relation was created. The owner of the relation is given
all privileges on that relation. The owner account holder can pass privileges
on any of the owned relations to other users by granting privileges to their
accounts. In SQL the following types of privilege can be granted on each
individual relation R:
 Select privilege
 Modify privilege
 References privilege
To create a view, the account must have SELECT privilege on all relations in
the view definition.
The mechanism of views is an important discretionary authorization
mechanism in its own right. It can be used to restrict access to particular
fields and tuples.
Privileges can be propagated using the GRANT OPTION, for example,

GRANT SELECT ON EMPLOYEE, DEPARTMENT TO A3 WITH

GRANT OPTION

User A3 can now propagate the privilege to other accounts by using

GRANT.

2
 Discretionary Access Control Based on Granting and Revoking
Privileges (Cont 1)

GRANT SELECT ON EMPLOYEE TO A4

does not give propagation right to A4.

REVOKE SELECT ON EMPLOYEE FROM A3

revokes select privilege on Employee from A3.

Limits on propagation of privileges can be implemented in two ways.
Limiting horizontal propagation to an integer number i means that an
account B given the GRANT OPTION can grant the privilege to at most i
other accounts. Vertical propagation is more complicated; it limits the depth
of the granting of privileges. Granting a privilege with a vertical propagation
of zero is equivalent to granting the privilege with no GRANT OPTION. If
account A grants a privilege to account B with the vertical propagation set to
an integer number j>0, this means that the account B has the GRANT
OPTION on that privilege, but B can grant the privilege to other accounts
only with vertical propagation less than j. In effect, vertical propagation
limits the sequence of GRANT OPTIONS that can be given from one
account to the next based on a single original grant of the privilege.

Mandatory Access Control

In addition to discretionary access control, in many applications, it is
necessary to classify data and users based on security classes. This is known
as mandatory access control. The commonly used model for multilevel
security, known as the Bell-LaPadula model, classifies each subject and
object into one of the security classifications such as TS (top-secret), S
(secret), C (confidential), or U (unclassified), with TS>=S>=C>=U. Two
restrictions are enforced on data access: no read-up (simple security
property), and no write-down (star property, *-property). These we have
considered earlier also.
To incorporate multilevel security into the relational database model, it is
common to consider attribute values and tuples as data objects. Hence, each
attribute A is associated with a classification attribute C in the schema, and
each attribute value in a tuple is associated with a corresponding security
classification. In addition, in some models, a tuple classification attribute TC
is added to the relation attributes to provide a classification for each tuple as
a whole. Hence, a multilevel relation schema R with n attributes would be
represented as

3
 Mandatory Access Control (Cont 1)

R(A1,C1,A2,C2,..,An,Cn,TC)

where each Ci represents the classification attribute associated with attribute

Ai.
The value of the TC attribute in each tuple t – which is the highest of all
attribute classification values within t – provides a general classification for
the tuple itself, whereas each Ci provides a finer security classification for
each attribute value within the tuple. The apparent key of a multilevel
relation is the set of attributes that would have formed the primary key in a
regular (single-level) relation. A multilevel relation will appear to contain
different data to subjects (users) with different clearance levels. In some
cases, it is possible to store a single tuple in the relation at a higher
classification level and produce the corresponding tuples at a lower-level
classification through a process known as filtering. In other cases, it is
necessary to store two or more classification tuples at different classification
levels with the same value for the apparent key. This leads to the concept of
poly-instantiation, where several tuples can have the same apparent key
value but have different attribute values for users at different classification
levels.
We illustrate these concepts with the simple example of a multilevel relation
shown in Figure 23.2 (a), where we display the classification attribute values
next to each attribute’s value. Assume that the Name attribute is the apparent
key, and consider the query

SELECT * FROM EMPLOYEE.

A user with security clearance S would see the same relation shown in
Figure 23.2(a), since all tuple classifications are less than or equal to S.

4
 Mandatory Access Control (Cont 2)

(a) Employee – the original tuples

Name Salary JobPerformance TC
Smith U 40000 C Fair S S
Brown C 80000 S Good C S
(b) Employee – after filtering for classification C users
Name Salary JobPerformance TC
Smith U 40000 C NULL C C
Brown C NULL C Good C C
(c) Employee - after filtering for classification U users
Name Salary JobPerformance TC
Smith U NULL U NULL U U
(d) Employee – poly-instantiation of the Smith tuple
Name Salary JobPerformance TC
Smith U 40000 C Fair S S
Smith U 40000 C Excellent C C
Brown C 80000 S Good C S
Figure 23.2 A multilevel relation to illustrate multilevel security
However, a user with security clearance C would not be allowed to see
values for salary of ‘Brown’ and JobPerformance of ‘Smith’, since they have
higher classification. The tuples would be filtered to appear as shown in
Figure 23.2(b), with Salary and JobPerformance appearing as null. For user
with security clearance U, the filtering allows only the Name attribute of
‘Smith’ to appear, with all other attributes appearing as null (Figure 23.2(c)).
Thus, filtering introduces null values for attribute values whose security
classification is higher than the user’s security clearance.
In general, the entity integrity rule for multilevel relations states that all
attributes that are the members of the apparent key must not be null and
must have the same security classification within each individual tuple.
Additionally, all other attribute values in the tuple must have a security
classification greater or equal to that of the apparent key. This constraint
ensures that a user can see the key if the user is permitted to see any part of
the tuple. Other integrity rules, called null integrity and inter-instance
integrity, informally ensures that if a tuple value at some security level can
be filtered (derived) from a higher-classified tuple, then it is sufficient to
store the higher classified tuple in the multilevel relation.

5
 Mandatory Access Control (Cont 3)

To illustrate poly-instantiation further, suppose that a user with security

clearance C tries to update the value of JobPerformance of ‘Smith’ in Figure
23.2 to ‘Excellent’; this corresponds to the following SQL update being
used:

UPDATE EMPLOYEE
SET JobPerformance=’Excellent’
WHERE Name=’Smith’

Since the view provided to users with security clearance C (see Figure
23.2(b)) permits such an update, the system should not reject it; otherwise,
the user could infer that some non-null value exists for the JobPerformance
attribute of ‘Smith’ rather than the null value that appears. This is an
example of inferring information through what is known as a covert channel,
which should not be permitted in highly secure systems. However, the user
should not be allowed to overwrite the existing value of JobPerformance at
the higher classification level. The solution is to create a poly-instantiation
for the ’Smith’ tuple at the lower classification level C, as shown in Figure
23.2(d). This is necessary since the new tuple cannot be filtered from the
existing tuple at classification S.
The basic update operations INSERT, DELETE, UPDATE must be modified
to handle this and similar situations, but we do not consider these issues
here.

Comparing Discretionary Access Control (DAC) and Mandatory

Access Control (MAC)
DAC policies are characterized by a high degree of flexibility, which makes
them suitable for a large variety of application domains. The main drawback
of DAC models is their vulnerability to malicious attacks, such as Trojan
horses embedded in application programs (we have already seen it
previously). MAC polices ensure a higher degree of protection – in a way,
that prevent any illegal flow of information. The drawback of MAC is that it
is too rigid: it requires a strict classification of subjects and objects into
security levels, and therefore, they are applicable to very few environments,
as military applications. In many practical situations, DAC is preferred
because it offers a better trade-off between security and applicability.

6
 Role-Based Access Control (RBAC)
RBAC emerged in 1990s as a proven technology for managing and
enforcing security in large-scale enterprise-wide systems. Its basic notion is
that permissions are associated with roles, and users are assigned to
appropriate roles. Roles can be created using the CREATE ROLE and
DESTROY ROLE commands. The GRANT and REVOKE commands
discussed previously can then be used to assign and revoke privileges from
roles.

Figure 1 is taken from [Osborn S., Sandhu R., Munawer Q., Configuring
role-based access control to enforce mandatory and discretionary access
control policies. – ACM Transactions on Information and System
Security, Vol. 3, No. 2, May 2000, 85-106,
http://cmpe.emu.edu.tr/chefranov/cmpe552-06/Lecture
%20Notes/OsbornACMTISS00.pdf ]

7
 Role-Based Access Control (Cont 1)

RBAC appears to be a viable alternative to traditional DAC and MAC. It

ensures that only authorized users are given access to certain data or
resources. Users create sessions during which they may activate a subset of
roles to which they belong. Each session can be assigned to many roles, but
it maps to one single user or a subject only. Many DBMS have allowed the
concept of roles, where privileges can be assigned to roles (e.g., Oracle).
Role hierarchy in RBAC is a natural way to organize roles to reflect the
organization’s lines of authority and responsibility. By convention, junior
roles at the bottom are connected to progressively senior roles as one moves
up hierarchy. The hierarchy diagrams are partial orders, so they are reflexive,
transitive, and anti-symmetric:

8
 Role-Based Access Control (Cont 2)
(x )( xx )
(x, y , z )( xy  yz  xz )
(x, y )( x  y  xy  yx)
Another important consideration in RBAC systems is the possible temporal
constraints that may exist on roles, such as the time and duration of role
activations, and triggering of a role by an activation of another role. Using
an RBAC model is a highly desirable goal for addressing the key security
requirements of Web-based applications. Roles can be assigned to workflow
tasks so that a user with any of the roles related to a task may be authorized
to execute it and may play a certain role for certain duration only.

Introduction to Statistical Database Security

Statistical databases are used mainly to produce statistics about various
populations. The database may contain confidential information about
individuals, which should be protected from user access. However, users are
allowed to retrieve statistical information about the populations, such as
averages, sums, counts, maximums, minimums, and standard deviations. We
will illustrate the problem with a simple example. Various techniques are
discussed, for example, in [Adam N.R., Wortmann J.C., Security-Control
Methods for Statistical Databases: A Comparative Study, - ACM
Computing Surveys, Vol. 21, No. 4, December 1989, 515-556,
http://cmpe.emu.edu.tr/chefranov/cmpe552-06/Lecture
%20Notes/AdamACMCS89.pdf ]
We use Person relation with a schema
Name Ssn Income Address City State Zip Sex Last
degree
A population is a set of tuples of a relation (table) that satisfy some
selection condition. Hence, each selection of the Person relation will specify
a particular population of Person tuples. Or example, the condition Sex=’M’
specifies the male population; the condition ((Sex=’F’) and
(Last_degree=’M.S.’ or Last_degree=’Ph.D’)) specifies the female
population that has an M.S. or Ph.D. degree as their highest degree; and the
condition City=’Houston’ specifies the population that live in Houston.
Statistical database security technique must prohibit the retrieval of
individual data. This can be achieved by prohibiting queries that retrieve
attribute values and by allowing only queries that involve statistical
aggregate functions. Such queries are sometimes called statistical queries.
It is the responsibility of a database management system to ensure the
confidentiality of information about individuals, while still providing useful

9
 Introduction to Statistical Database Security (Cont 2)

statistical information to users. Violation of privacy can be illustrated by the

following example.
In some cases it is possible to infer the values of individual tuples from a
sequence of statistical queries. This is particularly true when the conditions
result in a population consisting of a small number of tuples. As an
illustration, consider the following statistical queries:
Q1: SELECT COUNT(*) FROM PERSON
WHERE <condition>

Q2: SELECT AVG(Income) FROM PERSON

WHERE <condition>

Now suppose that we are interested in finding the salary of Jane Smith, and
we know that she has a Ph.D. degree and that she lives in the city of Bellaire,
Texas. We issue the statistical query Q1 with the following condition:
(Last_degree=’Ph.D.’ and Sex=’F’ and City=’Bellaire’ and State=’Texas’)

If we get a result of 1 for this query, we can issue Q2 with the same
condition and find the salary of Jane Smith. Even if the result of Q1on the
preceding condition is not 1 but is a small number – say 1 or 2 – we can
issue statistical queries using functions MIN, MAX, and AVG to identify the
possible range of values for the salary of Jane Smith.
The possibility of inferring individual information from statistical queries is
reduced if no statistical queries are permitted whenever the number of tuples
in the population specified by the selection condition falls below some
threshold. Another technique is to prohibit sequences of queries that refer
repeatedly to the same population of tuples. It is also possible to introduce
slight inaccuracies or noise into the results of statistical queries deliberately,
to make it difficult to deduce individual information from the results. Some
approaches to statistical database security are shown on the following figure
from Adam and Wortmann:

10
 Introduction to Statistical Database Security (Cont 3)

Flow Control
Flow control regulates the distribution or flow of information among
accessible objects. A flow between object X and object Y occurs when a
program reads values from X and writes values into Y. Flow controls check
that information contained in some objects does not flow explicitly or
implicitly into less protected objects. Thus, a user cannot get indirectly in Y
what he or she cannot get directly in X. Active flow control began in the
early 1970s. Most flow controls employ some concept of security class; the
transfer of information from a sender to receiver is allowed only if the
receiver’s security class is at least as privileged as the sender’s.

11
 Flow Control (Cont 1)

A flow policy specifies the channels along which information is allowed to

move. The simplest flow policy specifies just two classes of information:
confidential (C) and non-confidential (N), and allows all flows except those
from class C to class N. This policy can solve the confinement problem that
arises when a service program handles data such as customer information,
some of which may be confidential. For example, an income-tax computing
service might be allowed to retain a customer’s address and the bill for
services rendered, but not the customer’s income or deductions.
Access mechanisms are responsible for checking user’s authorizations for
resource access: only granted operations are executed. Flow controls can be
enforced b an extended access control mechanism, which involves assigning
a security class (usually called a clearance) to each running program. The
program is allowed to read a particular memory segment only if its class is
as low as that of the segment. This automatically ensures that no information
transmitted by the person can move from a higher to a lower class. For
example, a military program with a secret clearance can only read from
objects that are unclassified and confidential and can only write into objects
that are secret or top secret.
Covert channels can be classified into timing (temporal) and storage (spatial)
channels. One way to think of the difference between covert timing channels
and covert storage channels is that covert timing channels are essentially
memoryless, whereas covert storage channels are not. With a timing channel,
the information transmitted from the sender must be sensed by the receiver
immediately, or it will be lost. However, an error code indicating a full disk
which is exploited to create a storage channel may stay constant for an
indefinite amount of time, so a receiving process is not as constrained by
time.

Covert Channels in Networks

Further examples are taken from [Smeets M., Koot M., Research Report:
Covert Channels. - http://staff.science.uva.nl/~delaat/snb-2005-
2006/p27/report.pdf ]
In Figure 5, if a packet is received in the time window, then bit is 1,
otherwise, it is 0. There may be problems related with delivery delays.

12
 Covert Channels in Networks (Cont 1)

In Figure 6, packets may arrive not in the original order that results in
swapping of bits.
Covert channels using IPv4 are discussed below.

13
 Covert Channels in Networks (Cont 2)

Version = 4
Internet Header Length – Length of header in 32-bit words. The minimum
value is five, for a minimum header length of 20 octets
Type of Service – Provides guidance to end system IP modules and to
routers along the packet’s path, in terms of packet’s relative priority
Total Length – Total IP packet length, in octets
Identification – A sequence number that, together with the source address,
destination address, and user protocol, is intended to identify a packet
uniquely. Thus, the identifier should be unique for the packet’s source
address, destination address, and user protocol for the time during which the
packet will remain in the internet.
Flags – Only two of the bits are currently defined. When a packet is
fragmented, the More bit indicates whether this is the last fragment in the
original packet. The Don’t Fragment bit prohibits fragmentation when set.
Fragment Offset – Indicates where in the original packet this fragment
belongs, measured in 64-bit units. This implies that fragments other than the
last must contain a data field that is a multiple of 64 bits in length.
Time to Live – Specifies how long, in seconds, a packet is allowed to
remain in the internet. Every router that processes a packet must decrease it
by at least one, so the TTL is somewhat similar a hop count
Protocol – Indicates the higher level protocol, which is to receive the data
field at the destination; thus, this field identifies the type of the next header
in the packet after the IP header.

14
 Covert Channels in Networks (Cont 3)

Header Checksum – An error detecting check code applied to the header

only. Because some header fields may change during transit (e.g., time to
live), this is re-verified and recomputed at each router. For purposes of
computation, the checksum field is itself initialized to a value of zero.
Source Address – Coded to allow a variable allocation of bits to specify the
network and the end system attached to the specified network.
Destination Address – Same characteristics as source address
Options (variable) – Encodes the options requested by the sending user;
these may include security label, source routing, record routing, and time-
stamping.
Padding (variable) – Used to ensure that the packet header is a multiple of
32 bits in length

15
 Covert Channels in a Computer

Another examples are from

http://sconce.ics.uci.edu/ics243g/slides/lec14.pdf :

Example from http://www.list.gmu.edu/infs762/infs762su04ng/l1-access-

control.ppt :
Given 5MB pool of dynamically allocated memory
HIGH PROCESS
bit = 1 => request 5MB of memory
bit = 0 => request 0MB of memory
LOW PROCESS
request 5MB of memory
if allocated then bit = 0 otherwise bit = 1

16
 Covert Channels in Databases

Example is from http://ieeexplore.ieee.org/iel2/1114/7667/00318449.pdf?

arnumber=318449 :

17
Covert Channels in Databases (Cont 1)

18
 Audit
We follow [Hassan A. Afyouni, Database security and auditing,
Thomson, 2006, Chapter 7]
The auditing practice that is most often publicized is the review of an
organization’s documents such as the financial statements to make sure that
every change to the account is recorded and is accurate. An audit also
assures that all company transactions comply with government regulations
and laws. Most importantly, an audit can be conducted as a review of the
enforcement of security policies and procedures. All audits take place in an
auditing environment (database auditing environment), which includes:
 Objectives – An audit without a set of objectives is useless. Objectives
usually are set by the organization, industry standards, or government
regulations and laws
 Procedures – To conduct an audit, step-by-step instructions and tasks
must be documented ahead of time.
 People – Every auditing environment must have an auditor, even in
the case of automatic audit. Other people involved in the audit are
employees, managers, and anyone being audited
 Audit entities – This include people, document, processes, systems,
activities, or any operations that are being audited
 Database – Database activities, data changes, data accuracy, data files,
and operating system
Security measures are inseparable from auditing.

Auditing Process
The Auditing process (AP) ensures that the system is working and complies
with the policies, standards, regulations, or laws set by the organization,
industry, or government. It differs from Quality assurance (QA) and
Performance monitoring (PM). The QA process is active during the
development phase of the product and before implementation of the system.
It is aimed to assure that product is working correctly. The PM process is
active when the product is commissioned into production and is aimed to
monitor response time. The AP is active after product is commissioned into
production but is aimed on verification that the product or system complies
with policies, laws, and standards.

19
 Auditing Process (Cont 1)

The AP has the following phases:

In system development life cycle (SDLC), - when the system is planned,
analyzed, developed, tested, and implemented, - policies, laws, and industry
standards must be incorporated as part of the system requirements and
specifications. After the system is in production, the AP starts. The first
phase is to understand the objectives of the audit and plan a procedure to
execute the AP. In this phase you are identifying what should happen. The
second phase is to review, verify, and validate the system according to the
objectives set in the previous phase. Divergence from the auditing objectives
is documented. In this phase you are observing what is happening. The last
phase is to document the results of the audit and recommend changes to the
system. In this phase you are identifying the changes that are needed.

Auditing Objectives
Here are the top ten database auditing objectives:
 Data integrity
 Application users and roles – Ensure that user are assigned roles that
correspond to their responsibilities and duties
 Data confidentiality – Identify who can read data and what data can be
read
 Access control – Ensure that the application records time and duration
when a user logs onto the database or application
 Data changes – Create an audit trail of all data changes
 Data structure changes – Ensure that the database logs all data
structure changes
 Database or application availability – Record the number of
occurrences and duration of application or database shutdowns
(unavailability) and all startup times. Also, record all reasons for any
unavailability
 Change control – Ensure that a change control mechanism is
incorporated to track necessary and planned changes to the database
or application
 Physical access – Record the physical access to the application or the
database where the software and hardware resides
 Auditing reports – Ensure that reports are generated on demand or
automatically, showing all auditable activities

20
 Audit Classification
Internal – is conducted by a staff member of the company being audited
External – is conducted by a party outside the company that is being audited
Automatic – is prompted and performed automatically
Manual – is performed completely by humans
Hybrid – is a combination of automatic and manual audits

Audit Types
Financial, Security, Compliance (checks compliance with standards),
Operational (verifies that an operation is working according to the policies
of the company), Investigative (performed in response to an event, request,
threat, or incident to verify the integrity of the system), Product (to ensure
that the product complies with the industry standards), Preventive (to
identify problems before they occur).

Benefits and side effects of auditing

Benefits:
 Enforces company policies and government regulations and laws
 Lowers the incidence of security violations
 Identifies security gaps and vulnerabilities
 Provides and audit trail of activities
 Provides another means to observe and evaluate operations of the
audited entity
 Provides a sense or state of security and confidence in the audited
entity
 Identifies or removes doubts
 Makes the organization being audited more accountable
 Develops controls that can be used for purposes other than auditing

21
 Side effects
Frequent audits can cause the following:
 Performance problems due to preoccupation with the audit instead of
the normal work activities
 Generation of many reports and documents that may not be easily or
quickly disseminated
 Disruption to the operations of the audited entity
 Consumption of resources, and added costs from downtime
 Friction between operators and auditor
 From a database perspective, could degrade the performance of the
system; might also generate a massive number of logs, reports, and
data that require a system purge

22
 Auditing Models

If audited entity is registered in the auditing repository, the following are

recorded:
 State of the object before the action was taken along with the time of
the action
 Description of the action that was performed
 Name of the user who performed action

23
Auditing Models (Cont 1)

24
Auditing Models (Cont 2)

25
Auditing Models (Cont 3)

26