Detection Techniques for Fault Tolerance
Robert S. Hanmer · 2004
Errors must be detected before they can be recovered or tolerated in highly available or highly reliable systems. These patterns discuss a time-honored method of detection, a Watchdog timer, and its larger cousin the System Monitor. This mechanism functions by having one part of the system watch for status or “heartbeat ” messages to be delivered from the part being monitored. Other patterns help to make the monitoring and detection implementation efficient.