Wednesday, February 19, 2003 12:13 PM
JMX Agent to JMX Agent communication is generally done using a component of the JMX spec called a "connector". A connector can be implemented in many ways and protocols but the most prevelant at this time is RMI. Generally each JMX Agent has a set of connectors or adaptors which clients can use to connect and manage MBeans. So you have different options based on what the particular JMX implemenations provides. You can connect from one JMX agent to another using the connector or adapter directory (Generally adapters are used by lightweight clients like SNMP managers or web browsers). Or some implementations create proxy MBeans on each agent that represent remote MBeans in other agents.
In a simple case, you might have an MBean in JMX Agent A, which looks up another remote mBeanServer in JMX Agent B using JNDI or RMI lookup. It then can manipulate MBeans in that agent using the remote interface it retrieves.
With the proxy beans, everything looks local to the agent your in, and the request and response are marshalled to and from the requestor.