A Security Prioritized Computational Grid Scheduling Model - Cloud, Grid and High Performance Computing: Emerging Applications - page 97

Box 1. The SPMinMin scheduling algorithm

do until (T Complete != NULL) //there are more tasks
{
Create T high from T Complete // the tasks demanding high security
do until (T high , != NULL) // there are tasks in T high

{

for each task i from T high

{

Create N qualified,i

for each node j from node list N qualified,i

compute CT ij = ET ij + BT j

find the ECT(Earliest Completion Time)

for each task and its corresponding node.

Generate matrix ECT task(i),node.

}

from the matrix ECT, find the task with

minimum ECT =( ECT k,m ) ) // t th task on m th node

Schedule task t on node m,

Delete task t from T high and T complete

Modify BT m = BT m + CT km // begin time for the node m is

modified

}

}

4. High speed nodes are less secured and heavy
tasks require more security.
5. High speed nodes are less secured and heavy
tasks require less security.
6. High speed nodes are less secured and no
dependency between length of task and
security.
7. No dependency between speed of nodes and
security and no dependency between length
of task and their security requirement.

them varies from 8 to 100. The aim is to study
the performance of the two algorithms for differ-
ent job sizes (number of tasks) when the grid size
is kept constant. Figure 2a to 2g shows the simu-
lation results when the tasks to be scheduled on
a grid environment are fixed to 40 and the number
of nodes varies from 4 to 24. This is to study the
performance of the two algorithms for different
grid size keeping job size fixed.
Also, the experiment was conducted to observe
the time taken by both the algorithms under
similar environment. Conspicuous is the fact that
apart from offering better makespan the SPMin-
Min is better in terms of its own execution in
comparison to MinMin. Figure 3 shows the com-
parison over the speed of the two algorithms. It
has been observed that with the increase in the
number of tasks the execution time of the algorithm
improves exponentially.

The parameters needed for the simulation to
work in a secured grid environment are mentioned
in Table 1.
Figure 1a to 1g shows the simulation results
for the mentioned seven grid environments for
MinMin and SPMinMin when the number of
processing nodes for a grid environment is fixed
to 16 and the number of task to be scheduled on