Views within the WebSphere Real Time perspective

Each view within the WebSphere® Real Time perspective presents data in specific sections of the display.

The Controller window

The controller window provides the tools for you to select views of WebSphere Real Time data. There are two main tasks you can perform by using the controller window.
Manage custom views
You can create a customized view, and add it to the list of available views. For more information about creating customized views, see Customizing the WebSphere Real Time perspective.
Select different views
You can select different views, by using a combination box. The box is populated initially with predefined views. Customized views also appear in this box if a custom view definition file has been created and identified in the preferences.

All predefined views are identified by a System view: prefix. All customized views are identified by a Custom view: prefix.

The Plot window

This window displays event data as a simple plot graph. The X-axis of the graph shows the actual time when an event took place. The Y-axis shows the time taken for the event to occur. For convenience, the Y-axis values can be adjusted to display with a logarithmic scale.

Use the Threads window to control which threads are visible on the graph, and the color that is used to display each thread.

When you hover over data in the plot window, a window opens providing details of the trace point associated with the event.

A number of outlier events are highlighted by numbered circles on the graph. The number of outliers that is displayed is determined by the Number of outliers to analyze setting in the WebSphere Real Time preferences. Events are shown as highlighted outliers only if they fall outside the range of events that are used for the determinism score calculation. In other words, if the events take longer than twice the median value.

If a thread has a highlighted outlier event, the Analysis and Recommendations tab shows activities for that thread from all other views. Both system and custom views are included in this analysis. Use this analysis to help determine the cause of the outlier event. The following types of activity are reported:
  • The thread was delayed because of class loading.
  • The thread was delayed because all the available processors were busy with other threads, or if you are using WebSphere Real Time for RT Linux, higher priority threads.
  • The thread was blocked by a garbage collection activity.
  • Other activities from custom views. This report is most effective when you identify lower-level activity within a monitored operation, then create a custom view for that activity. For example, you might have a business method that uses socket writes, so you create a custom view to monitor socket writes. If any of the outliers in the business method are caused by socket writes, the time spent in socket write is provided. You can then create a composite view which includes the socket write custom view as the secondary view. The composite view shows how the socket writes behave in combination with other events.

The Summary window

This window displays various statistics, calculated from the data presented in the plot window. The statistics include:
  • Total events processed.
  • Maximum time taken.
  • Minimum time taken.
  • Mean value for time taken.
  • Median value for time taken.
  • The standard deviation.

The Threads window

This window lists the threads that are displayed in the Plot window. Use the check boxes to control which threads are visible in the plot graph.
Note: When you clear a check box next to a thread, that thread is also excluded from determinism calculations.
Click the color button next to a thread to change the color of that thread on the plot graph. The default colors, and the number of different colors, are defined in the Health Center preferences: File > Preferences... > IBM® Monitoring and Diagnostic Tools - Health Center > Displayers > Display colors. If the number of threads on the graph is greater than the number of colors specified in the preferences, some threads have the same color. Change the number of colors in the preferences for better color distribution, or change the color of individual threads by using the color button for each thread.

The Histogram window

This window provides an alternative display of data. It shows a histogram representation of the data displayed in the outlier plot window. For example, in the predefined class loading view, the histogram representation shows how many class loading events took 0 - 1 ms to complete, how many events took 1 - 2 ms to complete, and so on.

The Analysis and Recommendations window

This window displays the results of analyzing the collected data. The results are in the form of a determinism score. If the number of data samples is too low, the Health Center warns you that the determinism score might not be accurate. In particular, for Java™ method-based views, where the view descriptor might match multiple methods, a warning is displayed reporting that multiple methods have been matched.

The determinism score is calculated as follows:
  1. Select all the data points in the plot window.
  2. Calculate the median data point value - for example, the median time taken for a class loading event.
  3. Find how many events fall within the following ranges:
    • Median plus or minus 20% of the median value
    • Median plus or minus 40% of the median value
    • Median plus or minus 60% of the median value
    • Median plus or minus 80% of the median value
    • Median plus or minus 100% of the median value
  4. Calculate the average number of events for the ranges.
  5. The average number is the determinism score, expressed as a percentage.
The determinism score can be interpreted as shown in Table 1.
Table 1. Interpreting the meaning of a determinism score
Score Meaning
70 or less A very poor result. There is a wide distribution of results for the event, indicating uneven performance.
70 - 80 A poor result.
80 - 90 A good result.
90 or more A very good result. The results are distributed closely around the median value, indicating consistent performance.




© Copyright IBM Corporation 2007, 2016.
© Copyright 1997, 2016, Oracle and/or its affiliates.
US Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM Corp.