Node Performance Tuning

You can improve the performance of xMatters notification and application server nodes by:

The next sections describe these processes in detail.

Increasing the database connection pool

If you are deploying against a production-grade database such as Oracle 10g or SQL Server, you can tune xMatters for increased throughput by optimizing the connection pool configuration for your deployment.

Distributing Device Engine threads

You can optimize the throughput of your xMatters node deployment by configuring the Concurrent Threads settings of the Device Engines. By distributing threads based on the amount of notification traffic each protocol receives, you can balance the workload across the Device Engines that handle the most notifications. For example, if email is the most common notification method, allocate the most threads to your email Device Engines.

You can adjust the settings for the Device Engines within the xMatters web user interface.

Note: The Active Licenses control the maximum number of threads allowed, and the number of threads per Device Engine.

Adjusting maximum session size

You can help optimize the throughput of your xMatters node deployment by adjusting the maximum session size for Protocol Providers.

The session size represents the number of notifications transacted within a single session. For example, if you set the maximum session size of an email Protocol Provider to 20, and there are at least 20 notifications to dispatch, the Device Engine connects to the email server, sends 20 notifications, and then disconnects.

To tune the performance of Protocol Providers, ensure that the session size is as large as possible for each Protocol Provider in use. For email providers in particular, a session size of 20 to 50 is recommended.

Economical scripting practices

There are several ways an Administrator can manage the scripts to improve Event throughput. In particular, many of the out-of-the-box script packages contain external service requests and messages which may or may not be of value to a particular deployment.

Because the creation and dispatch of an external service request or message comes with significant overhead, any unnecessary external service requests or messages should be removed.

For example, in the default script the following situation is reported to the integration point:

$reportedFailure = @session::reportFailure($liveNotificationId,
"USER_HUNG_UP_AFTER_AUTHENTICATION")
@hangupEventMessage = @event::createExternalServiceMessage()
$hangupEventMessage.message = "Callout to " & $targetName & " ended
prematurely (USER_HUNG_UP_AFTER_AUTHENTICATION)."
@hangupEventMessage::send()
 

If this Event is not of interest, this portion of the script should be removed. For more information about scripting in xMatters, refer to the xMatters Online Developer's Guide.