Before praising Jolokia to be the fasted remote JMX solution on the world, it has to be confessed, that we don’t have any hard benchmarks yet. But this is something on the todo list.

Jolokia is especially a very fast JMX remote accessing solution for non-Java environments since for a client there is no need to start a JVM which needs some time to startup. E.g. most of the Nagios JMX plugins out there, require a local JVM installation and the plugins, which are implemented in Java, queries the MBeans via JSR-160 and prints out the results to standard output for a single check, opening a new connection for each request. For large Nagios installations which monitors hundreds of servers for several parameters this poses a performance problem. With Jolokia, it has been confirmed that it scales up very well in a Nagios environment with 500 servers monitoring about 10 parameters on each server every 5 minutes.

Especially bulk requests can speed up monitoring considerably. This unique feature, which is not available for JSR-160 connectors, allows for quering many parameters with a single server turnaround.

Jolokia uses own JSON (de)serialization library using a Lexer generated by JFlex. Before Jolokia 2.1.0, it was json-simple, but it is no longer maintained and doesn’t even use Java 5 generics). Along with the fact that JSON itself is a very compact data format with small overhead compared to RMI, this adds to the good performance experience.

In summary, all these facts indicate that the Jolokia is faster than standard JSR-160 connectors with communication of RMI. A benchmark is pending to put these statements on a solid ground.