Power cycling is the process of turning hardware off and then turning it on again. In the data center, technicians use power cycling to test the durability and reliability of network components. It is commonplace for servers to operate for months or even years without a reboot or disruption. These long periods of uninterrupted operation can place the entire system's resilience and reliability into doubt. Regular power cycling tests proactively monitor system performance and identify potential hardware failures arising from a hard restart. These tests can also reveal the amount of downtime that can occur after an unexpected hardware shutdown, and ensure proper system reconfiguration and accessibility after a restart. It's a best practice to make sure these tests comply with other disaster recovery and shutdown procedures.
Power cycling is often done remotely using systems management tools that stop/close applications, close any open data files, and then finally shut down power. This orderly shutdown process reduces the chances of accidental data or file corruption. Once the power is off, the system is restarted remotely, using technologies such as Wake-on-LAN. Technicians often wait several seconds before restarting the system to ensure that any dynamic volatile memory has cleared completely.
If a system has become unresponsive, or cannot be shut down remotely, technicians can manually cycle power by holding down the physical power switch for several seconds, waiting several more seconds for memory contents to clear, and then using the power switch to restart the system. This method, however, does risk some data loss if open applications or data files are not closed first.