Analytical modeling techniques abstract the features of a parallel system as a set of parameters or parametrized functions in order to make the modeling task tractable. This technique has been widely used in developing and modeling parallel systems. It is usually inexpensive and provides an abstract view of the hardware and software. However, the models are usually not accurate when compared to real executions due to simplifications in the modeling process. Also, the user is usually responsible for determining the model parameters, which requires detailed knowledge of the system and the modeling technique.
There are three kinds of analytical approaches that differ by how the models are expressed. Each of them is described below.