Where to start

This topic shows possible workflows and the needed steps to perform them.

So far, the introduction explained the buffer pool analysis and tuning tasks and what Buffer Pool Analyzer offers to solve these tasks. Table 1 summarizes possible workflows that guide you from a potential buffer pool related task through a sequence of activities to achieve your aim. For a more problem-oriented approach see also A generalized approach to performing analyses.

Table 1. Possible workflows for Buffer Pool Analyzer
If you want to Perform the following steps
Create several activity reports on the host.
  1. Collect performance data (or use data from a previous data collection). Refer to Collecting data and also to Determining what to collect. Decide about the type of report (summary or detail).
  2. Write a batch job that creates a report. See Creating activity reports and bpd files. Use the trace data file as input. Note that you can use the BPACTIVITY REPORT in the batch job without any options.
  3. Study the activity report in detail.
  4. Modify your batch job and use some of the BPACTIVITY options. Compare the effects in the reports.
  5. Use different input data (created with different collection parameters) and different BPACTIVITY options in your batch job and see how this effects the reports.

You will notice how data collection parameters (format, type, continuity, duration) relate to BPACTIVITY command options; above all, that data can only be reported if it was previously collected.
Learn about what kind of buffer pool information can be viewed on the client.
  1. Start reporting and select and open one of the sample bpd files. See Viewing performance data on the client.
  2. Expand the tree in the Reporting folder and study the contents in the different subfolders.
  3. View how comparison data is presented in graphical forms.
Get a first glance at the object placement or simulation capabilities on the client.
  1. Select and open a sample file. See Optimizing object placements and initial buffer pool sizes, respectively Simulating buffer pool behavior.

    You will notice when bpd files and trace data files are used.

  2. Start and follow the wizard. Accept the default input values.
  3. Note the purpose of each step, and the output from these functions.
Perform a simulation on the client with real performance data.
  1. Collect performance data through ISPF. Choose Short format, Detail data type, and collect data for approximately 10 minutes. See Collecting data and also to Determining what to collect.
  2. Note the options you have chosen in specifying data collection parameters. Note how many trace records were collected during the specified time.
  3. Download the trace data file to the client. See Downloading files from the host to the client.
  4. Start the simulation function on the client and select and open the trace data file you have created.
  5. Optional: To directly simulate the results of an object placement, right-click the object placement result file, then select Simulate...
  6. Repeat the simulations with different simulation parameters, object placements or both. Note how the parameters influence the execution times of simulations.
  7. Study the simulation results in more detail.
View performance data or perform an optimization of object placements and buffer pool sizings on the client with real performance data.
  1. Collect performance data (or use data from a previous data collection). Choose Short format, Detail data type, and collect data for a few minutes. See Collecting data and Determining what to collect.
  2. Write a batch job that creates a bpd file. See Creating activity reports and bpd files. Use the trace data file as input. Note that you can use the BPACTIVITY FILE command in the batch job without any options for default behavior.
  3. Download the bpd file to the client. See Downloading files from the host to the client.
  4. Start the view function or object placement function on the client and select and open the bpd file you have created.
  5. Study the optimization results in more detail. Note the recommendations. You might want to compare them with your actual system settings.

Consider creating a trace data file and a bpd file from every data collection, and keeping them together on the client. This lets you iteratively perform optimizations and simulations on the same snapshot of performance data.
Study a detailed use case. Read Example of a use case.