The need to protect storage arrays with battery backup, or an uninterruptible power supply, is a given. However, there are several important things to consider when selecting and configuring UPS hardware.
UPS Wattage Requirements
Selecting UPS hardware requires more than just matching the wattage of the storage array’s power supply to the wattage listed on a UPS. The main reason for this is that no UPS is 100% efficient. Every UPS is going to lose some energy in the DC-to-AC conversion process. As such, a 1,000-watt UPS might only be able to realistically provide 850 watts of output power. In this example, it would mean that the UPS is 85% efficient. In the real world, UPS hardware can be more or less efficient.
You never want to put yourself into a situation in which powering your backup hardware requires UPS hardware to operate at 100% capacity. Imagine that you have a 1,000-watt UPS that is 90% efficient. That would mean that 100% output for that UPS would be 900 watts. While the UPS should theoretically be able to supply 900 watts to a storage device without any problem, it’s entirely possible that the UPS won’t quite be able to deliver the full 900 watts. It’s also possible that your storage hardware might require a bit more power than you estimate. As such, it’s a good idea to leave yourself a margin of safety. I recommend a 15% safety margin between the UPS hardware’s theoretical output and the amount of power required by your storage hardware, but you could adjust that safety margin based on your comfort level.
UPS Hardware Compatibility Requirements
Hardware compatibility is another important consideration when selecting a UPS. While a UPS does need to be able to power your storage hardware if a power outage were to occur, the battery is only going to last for so long. As such, UPS hardware needs to be able to initiate a clean shutdown before the battery is completely depleted. To satisfy that requirement, the storage array needs to be able to communicate with the UPS. If you look at Figure 1, for example, you can see that my storage array automatically recognized the presence of a newly attached UPS.
The storage array was able to detect that a UPS had been attached.
A storage array that supports the use of a battery backup will almost always include settings that allow you to determine how long to leave the array running on battery backup before initiating a graceful shutdown. My array, for example, defaults to a value of 5 minutes.
In reality, the default shutdown period may or may not work for your hardware configuration.
For example, the battery backup that I am using is rated to run for 7.8 minutes at a 100% load of 900 watts. Of course, that assumes that the battery is fully charged, but for the sake of this article, we will assume that it is. However, this does not mean that the array will run for 7.8 minutes on battery. According to the manufacturer’s specs, my NAS is equipped with a 250-watt power supply and consumes 76.79 watts of power when fully loaded. That’s nowhere near the 900 watts that the UPS can supply. As such, the expected run time for the UPS would likely be over an hour. Incidentally, UPS hardware manufacturers post expected run times at various wattages.
UPS Hardware Shutdown Considerations
You also have to consider how long it takes UPS hardware to fully shut down. It can take several minutes for a storage array to shut down. However, the amount of time required to shut down can be longer if data is being held in a RAM cache because the cached data will need to be committed to disk before the shutdown can complete.
When deciding how long to allow a storage array to run on battery, it is important to consider not only the battery’s capacity, but how much time the array will need to shut down. It may also be advisable to add a couple of minutes to the anticipated shutdown time as a way of giving yourself an extra margin of safety.