There are two ways the Oracle Coherence Monitor (OCM) can communicate with a Coherence cluster. Based on which option you select, everything from system requirements and setup to command line options are different.

Option 1. Run the OCM remotely and gather grid data through a remote JMX port on one of your nodes.

Option 2. Run the OCM as a management node in your grid, removing the need for remote port access and improving performance ten fold.

For new users that want to get up and running quickly, Option 1 (remote JMX port) is recommended. If you decide later that you want to run the OCM as a management node, you can switch anytime.

Run via Remote JMX Port
In this configuration, the Oracle Coherence Monitor communicates with your cluster using the standard JMX remote port interface. In order to support this, you will need to configure one of the nodes in your cluster to run as a management node and JMX MBeanServer. You will also need to configure all of the nodes in your cluster that you want to monitor to expose their metrics to the management node. For details, refer to System Requirements and Setup or your Oracle Coherence documentation. As illustrated below, the management node you configure will collect metric information from all of the other nodes using TCMP and pass this information to the Oracle Coherence Monitor via a remote JMX port connection:

A remote JMX port connection is recommended if you:

• Are new to Enterprise RTView and want to get up and running quickly
• Have a small number of nodes and/or caches in your cluster (*see Performance Note below)
• Are able to expose a remote JMX port for at least one node in your cluster (NOTE: This node must be running on Java version 1.5.0+)
• Are unable to run the Oracle Coherence Monitor as a node in your cluster due to security limitations (e.g. no access to unicast or multicast ports on which Coherence is running, firewall issues, etc.)
• Want to limit interactivity (i.e. users can access information and reset statistics, but cannot delete objects)
• Want to secure your cache data (i.e. users cannot see the data in the cache, only the metrics)

Run as a Management Node
In this configuration, the Oracle Coherence Monitor runs as the management node in your cluster. In order to support this, you will need to configure all of the nodes in your cluster that you want to monitor to expose their metrics to the management node. For details, refer to System Requirements and Setup or your Oracle Coherence documentation. As illustrated below, the management node will collect metric information from all of the other nodes using TCMP and then will use an internal connection to JMX. This removes the need for remote port access and improves performance ten fold.

Running as a management node is recommended if you:

• Have a large number of nodes and/or caches (*see Performance Note below)
• Are unable to expose a remote JMX port for at least one node in your cluster
• Are able to run the Oracle Coherence Monitor as a node in your cluster
• Want to monitor cache data

Performance Note
A Coherence JMX node will create one uniquely named MBean of type Cache for each combination of Node and Cache in a cluster. If the total of number of Nodes times number of Caches is large (e.g. greater than 100), you may see performance degrade as there is significant overhead associated with aggregating metrics across a large number of MBeans and accessing it via remote port. When you run the Oracle Coherence Monitor as a management node, no remote port connection is needed.