Information Technology Reference
In-Depth Information
Algorithm 1.
Replica (Job)
{
On the selected cluster C
n
, for the submitted job J
j
do
Set the threshold failure rate λ
th.
Get the failure rates λ
lt
of each node on which allocation has been made
For each node, if λ
lt
> λ
th
do
{
Get all modules M
ij
allocated on node P
k
For each module
do
{
Randomly allocate it to any processor with λ
lt
> λ
th
}
Table 3. Job J
0
Module
(m
ij
)
Job Specialization
(J
j
)
Number of Instructions
(I
i
)
m
00
J
0
150
m
10
J
0
200
m
20
J
0
175
m
30
J
0
100
m
40
J
0
200
Table 4. Matrix B
ih0
for Job J
0
ILLUSTRATIVE EXAMPLE
m
00
m
10
m
20
m
30
m
40
To elaborate the working of the RBS an example
has been illustrated using one of the results of
the simulation study with other job as detailed
in Table 3 and Table 4. The parameters taken
are scaled down for the purpose of illustration.
All the data values are generated randomly and
conform to the similar studies. Table 5 and Table
6 represents the CT and the hamming distance
between nodes respectively for cluster C
0
, which
is the selected cluster on which the job has been
finally allocated by the TSM scheduler. Table 7
represents the processing time matrix for cluster
m
00
0
3
3
0
0
m
10
3
0
0
2
3
m
20
3
0
0
0
2
m
30
0
2
0
0
0
m
40
0
3
2
0
0
vious modules reallocated on other nodes to get
executed. The RBS thus works as a supplement
to the main scheduling algorithm by increasing
the clusters fault appetite.
Search WWH ::
Custom Search