Powering a Data Centre – why does the power fail even with a UPS & how to properly failsafe?
So you’ve gone and installed a UPS backing up your data centre. Excellent, you may think. Now I am fully power protected. But is that really the case?
Whilst it may be true that you are protected against spikes, brownouts and short duration power cuts, how long will your batteries last for? 5 minutes? 10 minutes? That will be fine for the common momentary power cuts but what happens if its an hour long power cut or an 8 hour power cut? Without a backup diesel generator, you won’t have sufficient backup. Your servers will just go down (or you will quickly shut them down safely before the battery autonomy shuts down the UPS).
With a backup generator, provided you have a big enough fuel tank (or top up the tank regularly), you can run for hours and hours without mains power.
When talking about failsafe systems, you will of course think one single UPS is one single point of failure and this is undoubtedly true. Whilst modern UPS are extremely reliable (especially when regularly tested and maintained), there is always a risk of it failing unexpectedly. Traditionally the best way to avoid this would have been to install a second (or even third) UPS in an N+1 / N+2 parallel arrangement.
In this situation, the load would be split between the online UPS but should one of them fail (or just be offline for servicing) the other UPS automatically takes on all the load. Whilst this is still correct and still common, the availability of modular UPS which contain multiples of smaller hot swappable power units is becoming more and more common.
The benefit of modular UPS’ is that not only does it give you redundancy at a cheaper cost (because you aren’t doubling up on a UPS but simply adding one or more additional power units over and above the peak load requirement) but the benefits of being able to swap out failed units easily and without specialist technical knowledge is a great plus.