ABSTRACT:
A real-time database system (RTDBS)
can be defined as a database system which is designed to provide real-time
response to the transactions of data-intensive applications,
such as banking,
stock market, and computer-integrated manufacturing.
It is difficult, in a RTDBS, to meet all timing constraints
due to the consistency requirements of the underlying database.
Efficient transaction scheduling algorithms are required
to maximize the number of transactions satisfying their timing constraints.
The algorithms should take the real-time
requirements of transactions into account in ordering data accesses,
while maintaining data consistency.
In this project, we concentrate on the development and evaluation of
transaction scheduling algorithms in RTDBSs.
We evaluate the real-time performance of existing and developed
transaction scheduling
algorithms using both analytic methods and the
simulation technique.
The relative performance of
the algorithms is evaluated in both centralized and distributed
RTDBS environments
in terms of fraction of satisfied timing constraints.
Another major contribution of this project is the investigation of
the performance impact of data replication
in distributed RTDBSs and development of efficient data allocation algorithms
for dynamic allocation of replicas to data sites
to obtain the best real-time performance.