In an embodiment, a method for monitoring elements of a distributed computing system is disclosed. In the embodiment, the method involves evaluating a metric event from a monitored element based on at least one of tags and conditions of the monitored element by applying declared group functions corresponding to declared groups over the metric event and at least one of admitting the monitored element into membership of a declared group, evicting the monitored element from membership of a declared group, and maintaining membership of the monitored element in a declared group based on the evaluation of the metric event.