Minimizing the Performance Impact of Running Background Services

As a system administrator, you can schedule, monitor, and distribute background services across all nodes in a PPM Server cluster. You can proactively monitor application performance to identify threats and eliminate them before system-wide performance problems or instability occur.

The following subset of PPM background services can be run in parallel to optimize performance and minimize bottlenecks:

  • Request Status Export Service

  • FX Rate Update Service

  • Evaluate TM Approvers Service

  • Task Actual Rollup Service.

  • Financial Metrics Update Service

  • Staffing Profile Period Sum Update Service

  • Project Planned Value Update Service

  • Resource Pool Rollup Service

These services can be run on the same or on separate server cluster nodes on two different entities, such as projects, work plans, time sheets, and so on.

Note: For information about how you can monitor background service activity, seeBackground services monitor.

Recommendations for Running Background Services

Keep in mind that the flexible service framework that you get with a clustered server environment comes with some risk. Use the following guidelines in enabling and scheduling the background services on your instance:

For medium to large PPM deployments, we recommend that you dedicate one PPM Server on a single JVM to processing PPM background services. This minimizes the impact that running services has on users, and enables you to better monitor background service performance. For optimal performance, set the SERVICES_ENABLED server configuration parameter to false to turn off services on nodes devoted to user traffic.

By default, nodes that run background services in a server cluster environment have one heavy-service consumer and two light-service consumers. This configuration reduces both memory consumption and CPU usage.

Try to schedule background services to run during periods of low activity, such as weekends and non-working hours.

Note: For Information about PPM background services and instructions on how to schedule them, see PPM background services.

Some of the services may run more often than necessary, while other services may need to be run more frequently. For PPM Servers devoted to services, schedule services to run only as often as necessary.

Your instance may be running more background services than you need.

Note: For Information about PPM background services and instructions on how to enable or disable them, see PPM background services.

Although you can easily run services across multiple nodes, we recommend that you start by running services on a single node. Later, when the workload calls for it, and after you determine that services are running correctly, you can add PPM Servers as dedicated services nodes (JVMs) that have the same configuration as the initial services node (one heavy service and two light services). It is always better to run with fewer services nodes and retain most processing capacity for end-user activity.

As always, test your solution to determine what is optimal for your environment.