Previous Table of Contents Next


TABLE 23.8 Measured Throughputs tor Scheduler Design Study

Experiment
No.
A B C D E TW TI TB

1 -1 -1 -1 -1 1 15.0 25.0 15.2
2 1 -1 -1 -1 -1 11.0 41.0 3.0
3 1 1 -1 -1 -1 25.0 36.0 21.0
4 1 1 -1 -1 1 10.0 15.7 8.6
5 -1 -1 1 -1 -1 14.0 63.9 7.5
6 1 -1 1 -1 1 10.0 13.2 7.5
7

-1

1

1

–1

1

28.0

36.3 20.2
8 1 1 1 -1 -1 11.0 23.0 3.0
9 -1 -1 -1 1 -1 14.0 66.1 6.4
10 1 -1 -1 1 1 10.0 9.1 8.4
11

-1

1

-1

1

1

27.0

34.6 15.7
12 1 1 -1 1 -1 11.0 23.0 3.0
13 -1 -1 1 1 1 14.0 26.0 12.0
14 1 -1 1 1 -1 11.0 38.0 2.0
15 -1 1 1 1 -1 25.0 35.0 17.2
16 1 1 1 1 1 11.0 22.0 2.0

Example 23.1 Consider the scheduler design study of Case Study 19.2. The purpose of the study was to find the best factor levels that maximize the throughput for three classes of jobs, namely, word processing, interactive data processing, and background data processing. The factor levels in the 16 experiments used in the 25-1 design and the throughputs obtained are listed again in Table 23.8. For simplicity, let us just consider the word processing jobs whose throughputs are listed in column labeled Tw. The maximum throughput is 28. The factor levels corresponding to this throughput have been circled in the table. The next highest throughput in the column is 27. The levels are circled in that row too. Notice that in both rows, factor A is at level -1, factor B is at level 1, and factor E is at level 1. The remaining two factors change their levels between these two values. One conclusion for this observation is that to get high throughput for word processing jobs, there should not be any preemption (A = -1), the time slice should be large (B = 1), and the fairness should be on (E = 1). The settings for queue assignment and requeueing do not matter.


Previous Table of Contents Next

Copyright © John Wiley & Sons, Inc.