Analyzing Root Causes

Before you perform this task, specify the time-series data (TSD) file that is to be used as the data source for the graph.

Performance Explorer’s automatic result-correlation feature facilitates root-cause analysis of network and server bottlenecks by correlating client-side issues with corresponding server-side measurements. Automatic result correlation identifies the server-side measurements that are most closely associated with specific client-side errors, thereby enabling you to better identify server-side problems and expedite debugging efforts. Result correlation also works in the reverse: Server-side issues can be correlated with client-side measures.

Automatic result correlation statistically correlates key measures with dependent measures. For example, if a significant increase in server response time is detected by a client-side measure at 18:20 (6:20 p.m.), automatic result correlation can identify the server-side measures that contributed to this drop in client-side performance.

  1. Click within the measurement graph that you want to analyze and drag your cursor to the right to select the time frame to analyze.
    Note: To slide the time line forward or backward in time, right-click the time line and drag your mouse right or left. The time line can also be moved vertically along its Y axis.
    Note: To select a shorter period of time for analysis, drag the time line to the right. To select a longer period of time for analysis, drag the time line to the left.
  2. Click Find Root Cause on the workflow bar. Alternative: Right-click in the graph and choose Root Cause Analysis. The Find Root Cause - Correlation Settings dialog box opens. The base measure appears in the Base Measure box.
  3. From the Correlate with list box, select the type of measurement you want to correlate with the base measurement.
  4. Adjust the date- and time-selection settings by selecting a new start date, start time, end date, or end time from the appropriate list box. Date and time selection settings are defined automatically based on the time frame selected in the time line.
  5. In the Results area, specify how to filter measure results based on how well they match. Choose one of the following options:
    • Click the Best [x] correlation numbers option button and specify the number of measures to be returned in the text box.
    • Click the Minimum correlation of [x]% option button and specify a minimum correlation relevance by typing a value in the text box.
  6. Click Next.
    Note: If you are correlating against client and server measures, client measures, or server measures, advance to step 10.
    The Correlation Measures Properties page opens.
  7. When correlating against a custom measure, you must define which measure groups are to be correlated. Select the groups you want to correlate against by checking the Measures check boxes.
  8. Optional: Add other measures by clicking Add File and browsing to and selecting a time-series data file. You can also remove measures by selecting them and clicking Remove File.
  9. Click Next to run the correlation. The Correlation Results page opens.

    By default, returned measures are listed in order of degree of correlation, and all correlations are selected.

  10. Click the Correlation Group and Name column headers to sort returned measures.
  11. Uncheck the check boxes for the measures that you do not want to include in the correlation graph.
  12. Click Finish to generate the correlation graph.