General Approach to Estimating Application Performance

Applies to AQTime 8.81, last modified on January 18, 2022

The sections below describe the general procedure of estimating application performance with AQTime.

General Approach

To estimate your application’s performance and find potential bottlenecks, use AQTime’s Performance profiler. Note that during profiling, AQTime collects a lot of statistics on each profiled routine or source code line, which may significantly decrease AQTime’s performance and slow down the profiling process. To estimate your application's performance faster and more effectively, first, you need to pinpoint potential bottlenecks (that is, the most frequently executed functions that take much time) and then perform a more detailed investigation of the application run and find parts of code that need to be rewritten.

Below is the general procedure of finding bottlenecks in your managed and unmanaged applications:

  1. Profile the entire application with the Performance profiler at routine level and get results.

  2. Analyze the profiling results. You will get a list of routines that take up the most execution time.

  3. Add the problematic routines to the line-level area and profile them with the Performance profiler.

  4. Analyze the profiling results and find the problematic code lines.

After you analyze the profiling results and find potential bottlenecks, you can rewrite the code snippet containing the problematic lines and then estimate the application performance anew.

Quick Analysis of Unmanaged Applications

You can quickly analyze the performance of your unmanaged applications with the Sampling profiler and determine what methods of your unmanaged applications tend to take up the most or the least execution time. The Sampling profiler works much faster than the Performance profiler, however, it collects less accurate data. To pinpoint the methods that you may need to rewrite, do the following:

  1. Profile the entire application at line level with the Sampling profiler and get profiling results.

  2. Analyze the results and find the most frequently used routines and lines.

  3. Add the desired routines to the line-level area and profile them with the Performance profiler to collect detailed and accurate statistics.

See Also

Best Practices and How to Tutorials
Performance Profiler - Overview
Sampling Profiler - Overview

Highlight search results