Differences in the way a network and a host operating system handle multicast traffic is used to create a passive monitoring construct that measures both multicast availability and isolation. In one aspect, a computer registers a multicast group with a network to multicast a simple heartbeat message; consumer computers look for this heartbeat to detect multicast failures. In another aspect, a heartbeat is generated on the same multicast group that is being monitored but on a different port than the application traffic for this multicast group. The network routes the heartbeat packets identically to those of the application data but the heartbeat packets are instead delivered to a monitoring process and not the end application on arrival at the consumer computer. A correlating computer serves as a stateful warehouse of all multicast publisher and consumer information using information obtained by client-side agents. Using information of who is publishing and who the consumers in the various parts of the network actually see, island detection and network failures can be detected.