Sitting in the ApacheCon session for Ganglia. Looks really cool, and might finally be the monitoring solution I'm looking for. It has a feature where the monitoring daemon will ping the data collection daemon if metrics change quickly, so that load spikes can be reported immediately (think collection interval every two minutes but you want to know about a load spike right away).
So far I've tried and given up on Nagios (too much of a pain to install and configure) and Hyperic (cool but I don't need a 500 mb monitoring daemon jvm on the machine I'm monitoring).
