Modern enterprise applications are demanding more and more memory. The hardware vendors made it possible to put in production powerful machines that have tens of gigabytes of RAM. While it's tempting to run a single JVM with a huge heap on such a big fat box, the side effect of the large heap is significant increase in GC times. In such configurations, the GC can reach tens of seconds during that the application may be completely unresponsive.
Cacheonix allows to create a cluster of smaller JVMs inside a server with large RAM. Because each JVM serves fewer users and uses less memory, the garbage collections occur at different times, so users exposure to GC delays becomes smaller and is spread in time.