Fig. 14 shows the number of miss counts in the job tracker when the
job tracker employs different scheduling algorithm. In Fig. 14, the
number of miss counts in the job tracker can be reduced by 150–300
when the tracker employs the proposed algorithm to be its scheduler.
In details, Fig. 15 shows the miss rate of the job tracker with different
scheduler running the middle workload. As Fig. 15 shows, the data
miss rate of the proposed algorithm can be improved by 10% compared
to First-in-First-out (FIFO) scheduler. Similarly, in the lightweight
workload, the proposed algorithm also reduces the data miss rate by
10%, as shown in Fig. 16.