Power failures are relatively straightforward to identify because an entire group of network devices will not be accessible. This is very easy to troubleshoot in person because the device lights are off and the fans are not spinning. However, if direct access to the data center is not possible and you have to investigate the issue from a remote site, power failure issues can become more difficult to troubleshoot.
There are a number of different things you should keep in mind when dealing with power outages:
- The source of the power in the building
- The power circuit’s load
- The level of redundancy for the building’s power supply system (depending on how critical the data center/devices are)
- The power capacity correlated to the devices’ power consumption
- The device power supplies’ proper functionality
- The device power supplies’ backup
If the environment is too complex, you should perform an audit (maybe with a licensed electrician) to find out the exact power sources and all the other parameters mentioned above to simplify the troubleshooting process for power-related issues.
If you have a situation where the power supply is failing or the power source is not very reliable, you may have problems where the device is intermittently and randomly available over a long period of time. This makes the troubleshooting process even more difficult.
One common way to prevent power failure problems is to ensure that an uninterruptible power supply (UPS) system has been installed, which will sustain the devices’ power requirements for a limited time until the main power system is restored. UPS systems can be scaled based on the protection needed. They can back up a single device, an entire rack, or an entire building, and they can run on batteries or by complex power generators.
With many enterprise-level network devices, you also have the option to install additional power supplies into the same chassis, meaning you can plug in a different power source into each power supply, thus ensuring complete power redundancy. If one of the power sources fails, the second one will be activated and the device will continue functioning at full power. This type of redundancy can ensure that any type of power source problem will not affect network functionality.
In order to have a predictable and manageable environment, you should always monitor how much power network resources are using. This can be done using special tools that provide measurements and statistics on different network locations. They can also provide interesting outputs, including peak power statistics for each device and possible oversubscription, which will help adapt the power supply system to the needs of the network.
By taking advantage of all the techniques presented above, you can ensure that the power going into the network devices will provide the maximum uptime.