|
|
|
|
|
|
|
Approximations and Estimates |
|
|
|
| |
|
|
|
|
The modeling of any system is only as good as the assumptions that underlie that model. In our memory models, we may assume at various times Poisson arrivals or uniform address distributionsassumptions which are only approximately accurate. Under such conditions, it makes little sense to create more complex and precise models when the underlying assumptions are only approximate themselves. Still, the analysis must be performed carefully. Interference that is negligible in one case may be quite significant with different modeling parameters. Each component of delay or contention must be carefully analyzed based upon the environment under consideration. If, based upon that analysis, one can show that a particular source of contention is not significant, so much the better; but we know this only after the analysis is done. |
|
|
|
|
|
|
|
|
|
Figure 6.24
The three basic parameters of simple queues. |
|
|
|
|
|
