Finally, we define performance, as soon discussed, as the number of steps needed by a methodology, an algorithm, to find a solution.
A step may be atomic, compound, or complex and be in any of the analysis, design, implementation, or testing planes. The performance measurement of counting steps assumes that on the average each step the same amount of effort. Since the performance is an order-of-magnitude measurement, knowledge of the detailed effort of each step is not necessary. This assumption is similar to the assumption that all instructions take equal time when doing algorithm performance analysis. Of course, a more detailed performance measurement could be defined where different weights can be used for different steps.
To achieve the best case performance, an environment is created that will allow the methodology to find a solution in the least number of steps. In a worst case performance, an environment is created that will prevent the methodology from finding a solution until the entire environment is visited. In an average case performance, environments are created for the methodology that requires a typical number of steps to find a feasible solution.