Monitoring and Alerting

For short running tests, CPU, disk, and network information is generally analyzed on completion of the tests. For long running tests or, more specifically, user acceptance or production systems, it will be necessary to issue alerts when CPU usage is too high, disks are too busy and so on.

There are a number of applications available on Linux to monitor the operating system, as well as JVMs and databases.

For example:

TIBCO Hawk^® http://www.tibco.com/multimedia/ds-hawk_tcm8-731.pdf
Nagios http://www.nagios.com/solutions/linux-monitoring
Zabbix http://www.zabbix.com/product.php
Cacti http://www.cacti.net/what_is_cacti.php

On Windows, the supplied Performance Monitor can be used to monitor the systems and add alerts if necessary. See the following articles for more information on using Performance Monitor:

Also, there is an agent available for Nagios that enables the combining of system information from different platforms.

You will only be able to determine the exact effects of some of the configuration options, and therefore the most suitable settings for these options on your system, after trying them out. If you are thinking of changing any of the options described, TIBCO therefore recommends that you run a test against the system to see how the Java Virtual Machine memory and the CPU usage, on both the database and the ActiveMatrix BPM machines, are affected by different values.