Figure 4.5 shows the mean response time as a function of the number of servers. In column (a), the high priority jobs have a smaller mean size, and in column (b), the low priority jobs have a smaller mean size. The differences in the number of servers preferred by each class, which we have discussed in the previous figures, can be read off of Figure 4.5 by observing that each of the plots in the figure have a ``U-shape'' and the bottom of the ``U'' indicates the optimal number of servers.
The key points made in Figure 4.5 are that: (i) the mean response time of both priority classes is sensitive to the number of servers and (ii) increasing the number of servers may reduce mean response time up to a point, but making the number of servers too large increases mean response time (the curve forms a ``U-shape''). This figure also reinforces the prior messages that the greater the variability of the high priority jobs, the greater the number of servers needed to mitigate this variability, and how prioritization is performed has a large impact on the optimal number of servers (the optimal number of servers is much larger in column (b) than in column (a)).