I might like the JVM to fail-fast rather than plod along in a bogged down state, but if I did, I wouldn't want it to wholesale crash (throwing an OutOfMemoryExc

I might like the JVM to fail-fast rather than plod along in a bogged down state, but if I did, I wouldn't want it to wholesale crash (throwing an OutOfMemoryException), but instead have my monitoring system start to throttle the work its doing and get other processes to help out (scaling). Indeed, you can do this; the JVM has simple facilities to report the memory use and heap status, which you can monitor and use to proactively scale around the bogged-down VM.