Basic Concept of Multilevel Database Security - Multilevel Security for Relational Databases - page 30

Table 2.14 Belief-Consistent Multilevel Secure
Relational Data Model

EMPLOYEE DEPARTMENT SALARY TC

Ahmed U Accounting U 7,000 U U

Mohamed UC-S Sales UC-S 10,000 UC-S UC-S

the database at his own security level and also can access the contents
of the database at lower levels. The user also has access to the beliefs
of users at lower levels. Users can define their beliefs through the new
verify mechanism. The users at each security level can decide what
information is accepted. The great advantage of the BCMLS model is
the fact that the accepted information does not need to be replicated
or borrowed.

2.5 Performance Study

This section describes the performance study of relational multilevel
database models—SeaView, Jajodia-Sandhu, Smith-Winslett, MLR,
and belief-consistent models—and illustrates the impact of varying
the size and structure of the relational multilevel database on the per-
formance of these models.
The machine that is used for the implementation of the perfor-
mance study consists of CPU speed of 2.2 GHz, physical RAM size
of 3 GB, and hard disk size of 320 GB. The software that is used in
the implementation is Microsoft SQL server 2008 R2 and the experi-
ments' measurements were captured at the machine using a moni-
toring tool provided by the Microsoft SQL server. The experiments
measure the impact of changing the number of tuples, the number of
attributes, and the number of security levels on the performance of
the relational multilevel database models. These experiments define
the CPU response time (in seconds) as metric. For each query, the
monitoring tool observes the time that is taken for the system to give
the result of the query.

2.5.1 Experimental Database Structure

An experimental database, the timesheet database, consisting of four
relations, was created and populated to facilitate the performance
study. The employee relation provides information about employees,