71Configuring and Running a Simulation
common characteristics such as the protocol. However, even if a protocol was explic-
itly congured for a simulation study, it may not be a good idea to collect all the
available statistics of the particular protocol.
The next choice is equally obvious and deals with statistics that help to answer
the simulation study questions. For example, if the goal of the simulation is to
examine the inuence of an FTP application on the performance of the network
layer, then the appropriate network layer statistics should be collected even if
the simulation employs only default conguration values of the network layer
protocols.
Finally, it is often a good idea also to select statistics that are not directly related ...