You get the best performance predictions for a system from real-world data from a variety of workloads. For many reasons, robust, real-world data might not be available (see “Computation of performance predictions”). This section mentions some of the methods that Storage Modeller uses to improve its performance predictions when real-world data is absent or not robust.

Storage Modeller's algorithms for mixed queues of workloads improve performance reports. For example, the algorithms improve the Response Times plot (see “Interpreting the response-time curve”). These algorithms dynamically account for the characteristics of your design, including these factors:

Mixed-queue modeling is one reason that performance assessments in Storage Modeller can become more accurate over time. Paradoxically, the results of a performance assessment can sometimes decrease as the accuracy of assessment increases.