A Security Prioritized Computational Grid Scheduling Model - Cloud, Grid and High Performance Computing: Emerging Applications - page 95

site. Merely imposing security demand
in the MinMin degrades its performance
which is discussed through cases in the
later sections.

turn can be a single processor or a group of het-
erogeneous or homogeneous processors. A grid
job is comprised of “n” independent tasks. The
aim is to find an optimum schedule for assigning
the grid job (all the tasks) to the processing nodes
that satisfies the security constraint. Following is
the list of terminologies, used in this article.
T complete is the list of all the task of the given
job that is to be scheduled.
T high is the list of all the tasks with highest
security requirement.
A task is characterized as T i = {L i , SD i } where,
Li is the length (size) of the task (number of
instructions in the task), and SD i is the security
level demand of the task.
A processing node is characterized as N j = (SP j ,
SL j , BT j ) where, SP j is the execution speed of the
processing node, SL j is the maximum security
level offered by the processing node, and BT j is
the begin time of the node (time to execute the
tasks already assigned to the node).
N qualified,i is the list of processing nodes on which
the i th task can be executed i.e. list of the nodes
meeting the security demand of the i th task.
A schedule of the job is a set of n triplets <P j ,
T i , CT ij > where, P j is j th processing node, T i is
i th task, CT ij is completion time of i th task on j th
processing node.
CT ij = ET ij + BT j where, ET ij is execution time
of i th task on j th processing node, and BT j is the
begin time of j th processor.
Earliest Completion Time (ECT) of a task is
the minimum time amongst all the selected nodes
taken to complete the task.

PROPOSED MODEL

This work proposes a guaranteed security aware
scheduling model as Security Prioritized MinMin
[SPMinMin]. It is a security aware scheduling
model and assures security requirement of the
job (multiple tasks) unlike algorithms like Suffer,
DFPLTF, MinMin, MaxMin etc. MinMin gives
highest priority to the task that can be completed
first. Song et al. (2005) secured the MinMin by
merely imposing security restrictions on it. It
degrades the performance as security requirement
act as the limiting factor rather than guiding fac-
tor. This work modified the Min-Min algorithm
where the security demand of the task is the
major guiding factor for scheduling decisions.
SPMinMin allocates highest security demanding
tasks first on faster resources. Tasks having same
security requirement are then scheduled according
to MinMin. Thus it never compromises with the
benefits of MinMin but simply modifies it to work
efficiently in a secured environment. Extensive
experiments have been conducted over simulated
grid environments for both MinMin and the SP-
MinMin. The experimental test bed is divided on
the basis of possible heterogeneous scenario that
may exist in a real grid environment. The results
obtained clearly indicates that the proposed model
performs better and at the same time proves that
in non-grid environment MinMin can outperform
SPMinMin The results also reveal significant
performance gain and better site utilization of
SPMinMin over MinMin.

SPMinMin

In a Grid, tasks with different levels of security
requests compete for the resources. It is assumed
that a task with low security requirement can be
executed on both types of resources; the one of-
fering high security as well as the one offering
low security. Thus, a task with a desired security
level service can be executed only on a resource

Terminologies Used

Grid is considered to be composed of number of
non dedicated processing nodes and the node in