Ganglia is a massively scalable cluster monitoring and execution environment written at the University of California, Berkeley Computer Science division as part of the Millennium Project. Ganglia monitors over 25 core host metrics (such as CPU, load and memory) and any number of custom metrics on clusters with hundreds of nodes. Clients can plug into the Ganglia monitoring core via an XML interface to generate a multitude of useful visualizations. rrdtool is the tool of choice for Web visualization of cluster status and historical trends.

Ganglia Project page...