Abstract
M.Sc. (Computer Science)
Concurrency control is the problem that exists in a database
management system when more than one transaction or application
is executed simultaneously. If transactions or applications are
executed sequentially there will- be no problem with the allocation
of resources. It is however necessary to execute transactions
concurrently to utilise computer and resource capacity to
its maximum extent. It can lead to inconsistent data if this
concurrent execution of transactions are not properly controlled.
If this should happen the data would be of no more use
to the users of a system.
The thesis is divided in the following way. Chapter 1 gives
background information on the concurrency control problem. In
chapter 2 a couple of mechanisms for solving the concurrency
control problem are studied briefly. Chapters 3 and 4 provides a
more in depth study of two specific mechanisms namely two-phase
locking and timestamps. 80th of these mechanisms have already
been implemented in systems to. solve the concurrency control
problem.- In chapter 5 a comparison is made of the two methods
described in chapters 3 and 4. A third method for handling concurrency
control is briefly described in chapter 6. This method
hasn't received a lot of attention from researchers yet. And in
the last chapter, chapter 7, the concurrency control method used
in the SDD-1 system is studied in more detail. SDD-1 is a distributed
database management system.