Server configuration

Dedicated server

The user experience might suffer if you use the same server to process large volumes of software catalog data and to interact with the user interface. You can schedule imports for off-hours, or you can use a different server to process data for imports. Using a separate server also makes it easier to tune settings for performance because you can restart the server without disrupting user activity.

The server dedicated to processing the data for imports generally connects to the same database as the servers providing the user interface. However, you can use the mxe.crontask.donotrun parameter settings to isolate running the cron task to the dedicated data processing server. The instructions mention clustered environments, but they also pertain to an environment with separate servers connected to a single database.

If the server services more than one sizeable integration, it can be helpful for administration purposes to configure queues for each integration.

Application server memory

A message is created in the queue for each record that comes from the software catalog, so the queue itself expands to several hundred MB. To avoid out-of-memory errors, allocate a minimum of 1-GB heap space to the Java™ virtual machine for the application server. In some JVMs, there can also be a dependency on the MaxPermSize parameter, requiring a minimum value of 256 MB. Startup parameters from a successful server implementation include the following switches: -Xms512m -Xmx1028m -XX:MaxPermSize=256m.

For large software catalogs, consider configuring paging for queue messages as a way to manage memory requirements. For detailed information, see the documentation provided with your application server.