What is important when monitoring IBM Business Process Manager at the infrastructure level?

Again, this popped up in my Twitter stream of consciousness, thanks to @IBM_BPM : -

In my customer engagements, I am often asked the following questions:

• What metrics should we use to monitor the infrastructure for our IBM Business Process Manager  environment?

• What key metrics should I always look at to prevent the environment from becoming unavailable?

• Which actions should be taken in case the metric threshold is exceeded or when receiving an alert by running out of service?

For monitoring to work, agents must be installed on the server that will communicate with a tool for generating alerts and record values. Usually the client already has the tools for monitoring the infrastructure but they do not know what metrics are important to collect for IBM Business Process Manager, what the thresholds are, and what action to take in case an incident happens. Therefore, you need the IBM professional at the customer site to provide this kind of information and so you can customize your dashboards with the information provided.

Normally, because the IBM Business Process Manager is based on the WebSphere Application Server, we always get the same metrics as used on the application server. To ensure the smooth operation of servers and anticipate potential problems, monitoring is important.

What is important when monitoring IBM Business Process Manager at the infrastructure level?