This course is run at the The University of Nottingham within the School of Computer Science & IT. The course is run by Graham Kendall (EMAIL : gxk@cs.nott.ac.uk)
There is probably no such thing as an average process but studies have been
done on typical processes. It has been found that processes come in two varieties.
Processes which are I/O bound, and only require the CPU in short bursts. Then
there are processes that require the CPU for long bursts. Of course, a process
can combine these two attributes. At the start of its processing it is I/O bound
so that it only requires short bursts of the CPU. Later in its processing, it
is heavily I/O bound so that (if it was allowed) it would like the CPU for long
periods.
An important aspect, when scheduling is to consider the CPU Bursts that are
required. That is how long the process needs the CPU before it will either finish
or move to a blocked state. From a scheduling point of view we need not concern
ourselves with processes that are waiting for I/O. As far as the scheduler is
concerned, this is a good thing, as it is one less process to worry about.
However, the scheduler needs to be concerned about the burst time. If a process
has a long burst time it may need to have access to the CPU on a number of occasions
in order to complete its burst sequence.
The problem is that the scheduler cannot know what burst time a process has
before it schedules the process. Therefore, when we look at the scheduling algorithms
below, we can only look at the effect of the burst time and the effect it has
on the average running time of the processes (it is usual to measure the effect
of a scheduling policy using average figures).
Last Page | Back to Main Index | Next Page |
Last Updated : 13/01/2002