Skip to main content

The Performance Analyzer (PAL)

The Performance Analyzer (PAL) provides a view of all the performance statistics that Pega Platform™ captures. Use PAL to understand the system resources consumed by processing a single requestor session.

PAL is available on the Performance landing page (Dev Studio > Configure > System > Performance) or from the Performance tool in the toolbar. The following image shows how to access the PAL from the Performance tool in the toolbar:

Accessing PAL from the Performance tool in the toolbar

You can also use PAL to debug Constellation applications directly from a user Web Portal, as shown in the following image:

Accessing PAL from the web portal

Measurement of application performance

The first step to measuring your application performance is to take measurements. Start by clicking Reset Data to clear any data in the tool. Because the system continuously monitors performance, you are eliminating any previously recorded entries from your results by resetting data.

To add a simple reading, click Add Reading. When adding a reading, the best practice is to define points that identify what occurred during that reading. For example, use one reading per flow action or screen render, depending on what process you are measuring.

Tip:  Check PAL when the interaction or problem occurs to avoid accumulating a long list of calls that can make analysis more complex.

You can download the results to an Excel file by clicking Save Data.

Note: To learn more about the system utilization for a requestor session with Performance Analyzer, see Analyzing system performance.

Check your knowledge with the following interaction:

Summary display

The INIT row displays the totals from the first time the Performance tool made a reading. Each subsequent reading is labeled as a DELTA, which indicates the change from a previous reading. The TOTAL reading is the total sum of all the statistics from the last time the data was reset.

Match the numbers to the following image to learn more about the PAL summary display:

1. Total: Statistics in the top row reflect the totals since that start of your session or the most recent Reset Data operation.

2. Delta: Delta rows are displayed as you interact with Pega Platform. Each row shows values that accumulated from the previous Delta or Init row.

3. Init: The bottom row shows the totals since the first intitialization of the Performance tool. Most readings in this row at the outset are near zero. If the readings are not near zero, click Reset Data to reset the readings.

PAL summary display

Analysis of performance data

When you click INIT, DELTA, or TOTAL in the summary display, PAL opens a detailed breakdown of performance data for that reading type. PAL organizes this data into two categories to help pinpoint performance issues:

  • DX API sessions: Data from DX API calls that handle modern web interfaces and external system integrations
  • Studio interactions: Data from activities within Dev Studio, such as rule editing and administrative tasks

The following figure shows how PAL organizes performance data into DX API sessions and Studio interactions:

PAL data divided into DX API sessions and Studio interactions

When you select an individual DX API session or Studio interactions, PAL displays a detailed performance snapshot with specific metrics including resource consumption, database interactions, and processing details.

Match the numbers to the following image to learn more about the PAL Snapshot display:

1. Int #: Sequence number of the most recent completed interaction.

2. Int Count: Total number of interactions.

3. Total Elapsed: Total elapsed time for this interaction.

4. RA Elapsed: Rules assembly elapsed time.

5. Rule I/O Elapsed: Elapsed time spent finding rules.

6. RDB I/O Elapsed: Elapsed time executing SQL operations against the PegaRULES database; excludes Connect SQL executions for the PegaRULES database or external databases.

7. Connect Elapsed: Time spent waiting for an Connect rule execution to an external system.

8. Other I/O Elapsed: Elapsed time accessing non-rule-resolved instances from the PegaRULES database plus the elapsed time retrieving non-rule database lists.

9. Total CPU: Total CPU time used.

10. RA CPU: CPU time spent assembling Java classes.

PAL Snapshot display

 

In the two following images, which represent a single reading, the top delta has a reading of 6.42 for RA Elapsed. All values are in seconds. RA Elapsed represents the time spent in rule assembly. These results can skew performance readings as rule assembly, also known as first use assembly (FUA), is expensive and only occurs once. This is evidenced by the following results. The total elapsed time was 11.34 seconds, with 6.42 seconds of that time spent in rule assembly. Without the additional 6.42 seconds, the total time would be less than half the measured number. FUA also affects other readings such as the total rules run, the reads from the database, and various I/O counts.

PAL single reading - 1
PAL single reading - 2

To obtain results unaffected by FUA, run through the process once to ensure that all rules are assembled before taking any measurements. In the previous example, a run-through of the process has not occurred to demonstrate the impact on performance readings.

The readings are grouped into three different types that are signified by the following keywords in their labels:
  • Elapsed is the system time, in seconds, that a process takes; this time includes the CPU processing time. The time is generally equal to or longer than the CPU time. For example, Elapsed time compiling rules measures the amount of elapsed (system) time the system takes to compile the generated Java code for a rule.
  • CPU is the amount of CPU processing time for the current action in seconds. For example, CPU time compiling rules measures the amount of CPU time the system takes to compile the generated Java code for a rule.
  • Count is a count of the number of times a specific action occurs. For example, the Activity Count gives the number of activities started by a thread.

A result of 10 minutes may be acceptable in one situation, where a result over 100 milliseconds is considered too slow in another. Work with the lead system architect and the business architect to determine an acceptable result for each step of the process.

Check your knowledge with the following interaction:


This Topic is available in the following Module:

If you are having problems with your training, please review the Pega Academy Support FAQs.

Did you find this content helpful?

Want to help us improve this content?

We'd prefer it if you saw us at our best.

Pega Academy has detected you are using a browser which may prevent you from experiencing the site as intended. To improve your experience, please update your browser.

Close Deprecation Notice