User-facing bottlenecks and performance issues

The Improvement Plan is often the number one tool for PDC users because it shows the most frequent and worst-performing items in several useful categories. Although the Event Viewer gives you a comprehensive and detailed view of issues from the past 14 days, the Improvement Plan provides a more directed view of the top issues that you can resolve.

Improvement Plan: TOP5

The Improvement Plan - TOP5 landing page displays the top five cases for each category. You can choose to see the cases in each category with the biggest impact based on the overall impact, web interactions, service interactions, or background processing.

PDC organizes cases in the following categories:

• Database: Issues that occur in your database configuration, such as the slowest performing database calls.
• Connectors: Issues with connecting to outside systems from your application. This category shows the slowest connector invocations.
• General Performance: Slow processing time for various internal components, such as queue processors, job schedulers, and rule assemblies.
• Stability: Errors, exceptions, and run-time issues that affect overall system stability.
• Runtime guardrail: Alerts that inform you when an application seems to be violating best practices, such as an inefficient use of data pages by processing too many items in a single step or by processing items that are too big.
• Custom cases: Custom alerts that you define.
• Decisioning: Alerts from Pega Decision Strategy Manager and Pega Customer Decision Hub™ applications.

You can use the Application drop-down list to view only the issues that originated from a single application. You can then focus on your application. However, the list might not include some issues, such as certain system exceptions, if the system cannot determine the context.

In the following image, click the + icons to review the controls for the Improvement Plan - TOP5 landing page:

Every case in the Improvement Plan represents a unique problem that PDC identifies, a performance bottleneck in an application or your system.

To prioritize your cases, look at the following details:

1. Impact. Consider the business impact. Focus on cases that are highly visible or business-critical.
2. Part of the application that is affected. Learn how the case corresponds to specific actions or data processing in the application, such as information that the system displays, a feature, or background processing.
3. Expected user experience. Sort the area of impact by type:
   • By Web: Web cases are usually the most critical. Slow requests from web browsers mean that end users are blocked and waiting for data to load.
   • By Service: Service cases block users when synchronous services deliver data, which happens in real time.
   • By Background: Background impact cases involve data that queue processors, job schedulers, and agents process asynchronously in the background. These cases are not always as critical as web or service cases.

Case details

In the following image, click the + icons to see some of the case details that are often important:

The Slow Interactions landing page shows all cases for slow client-facing interactions:

• The PEGA0001 browser interaction events, which are single HTTP-based server interactions that exceed 1 second
• The PEGA0069 client rendering events, which capture the page rendering time on the client browser (including multiple server interactions) that exceed 4 seconds
• The PEGA0011 service interaction events, which capture service request processing that exceeds 1 second

This landing page provides another method for locating performance issues.

In the following image, click the + icons to see some of the features of the Slow Interactions landing page:

The Interactions and Interactions Per Node landing pages provide usage metrics for how your customers interact with applications over time. The Interactions landing page shows combined application-specific information for all nodes. The Interactions Per Node landing page displays more detailed information for each node.

Usage metrics help you to stay informed about the performance of your system by comparing the traffic across different requestors and nodes. With this information, you can check how busy your system is and what kind of traffic the system processes most often, and then assess the performance of your system so that you can respond to ongoing and potential issues. By analyzing this data, you can identify the patterns and sources of processing demand, which is helpful in performance tuning.

The Interactions landing page provides the high-level overview that you usually require. You can view all applications or select an individual application. The Interactions Per Node landing page helps to assess how traffic is distributed across the nodes in a cluster, and how each node responds under the load. This data is helpful for system administrators who look at metrics from a node perspective; however, for the majority of users, the simpler Interactions landing page provides all the necessary data.

In the following image, click the + icons to review the controls for the Interactions landing page:

Improve your Pega Platform applications by troubleshooting user-facing bottlenecks in PDC. You can use the following process to locate the issues that have the biggest impact:

1. Pick a TOP5 case, starting with the biggest web impact.
2. Identify the part of the application that the case affects the most and prioritize the case based on business impact.
3. Look for opportunities to optimize the SQL statement, indexes, or the underlying data model.

To view a 15-minute demonstration of the troubleshooting tools, see Using PDC to address performance bottlenecks.