Figure 2-4. Sample instrumentation analysis code and the actual results. After repeated product runs, clearly module C must wait for IO reads to complete.

The development manager quickly realized that he could also summarize the instrumentation by the resource, not only by the module. As shown in Figure 2-5, the development manager was clearly able to see that both IO reads and CPU consumption have a dramatic effect on the overall application performance.

Figure 2-5. Sample instrumentation analysis code and the actual results focusing on the resources, rather than the modules. After repeated product runs, waiting for IO reads to complete is the overall performance-limiting factor.