From the blog.

Managing Digital Racket
The more I tune out, the less I miss it. But that has presented me with some complex choices for a nuanced approach to curb

Street Power vs. UPS Power In Redundant Power Supply Devices

619 Words. Plan about 4 minute(s) to read this.

A question came into the Packet Pushers mailbox along these lines.

If a unit has two power inputs, should one go to UPS and one to street power, or is it better to have both power supplies fed by the same UPS?

The issue with raw street power is that it isn’t conditioned. So, voltage and amperage irregularities might happen — events affecting the power grid can also affect whatever is plugged into it. One visible effect of this is a brown-out, where lights dim when the power grid is overtaxed. Voltage spikes can also occur, often in the form of nearby lightning strikes, although lightning can find all sorts of paths into a device, not necessarily following power lines. In any case, this sort of unpredictability in the street electrical supply can damage power supplies in IT gear, making conditioning of some sort important.

One issue with in-rack battery-backup UPS gear is that over time, a UPS will fail. The failure could be catastrophic, resulting in a complete power failure, although it’s far more likely that the battery will simply cease to hold a charge, rendering the UPS a very heavy power strip. UPS devices have a lifespan of a few years. Eventually, the UPS will require complete replacement or partial replacement if the battery is serviceable. In addition, intelligent UPS management modules can lose their little minds, resulting in shutdown of power ports in certain models. (Anything “smart” can outsmart itself.)

Another in-rack UPS issue worth considering is that in-rack models don’t do well when overloaded (i.e. more electrical draw is requested than they are rated for) meaning that they’ll overheat or otherwise shutdown if too much is plugged in. You can’t power the Space Shuttle with a 9-volt battery. Even when a UPS is working properly and loaded appropriately, it will likely only offer minutes to an hour or so of uptime during a total power outage, although that depends on just how much load is placed on it by the plugged in equipment.

The mighty Packet Pushers labs feature 2 UPS's to handle the load during peak testing periods. (Yes, really.)

The Packet Pushers lab feature 2 UPS’s to handle the load during peak testing periods. (Yes, really.)

The point of the above paragraphs is less to highlight the shortcomings of various power sources and more to point out that you don’t want your power supply eggs in a single power source basket. While power problems are rare, they happen. The rule I follow is power diversity, if at all possible. Don’t make the mistake of running dual power supplies off of the same power source. Yes, there are risks with UPS’s, and there are risks with street power, so hedge your bets. Here’s a strategy that I believe is a good compromise.

Use street power for one supply,
conditioned by a quality surge suppression strip.

Use the UPS for the other power supply.

Many power strips are rack-mountable, so you should be able to keep things neat and tidy, as opposed to dropping a power strip on the floor under the rack. This all depends on budget.

A final point worth considering is that you don’t want to rely on street power and in-rack UPS if you don’t have to. In data center deployments of size, a common practice is to have an “A” power feed and a “B” power feed (and perhaps more) that provide fully conditioned and redundant power feeds throughout the data center. In these scenarios, the data center is powered by street power which is conditioned and fed into power distribution units throughout the DC. When the street power fails, a generator kicks on to power the DC. A bank of batteries buffers the quick transition between street power and generator.

Ethan Banks writes & podcasts about IT, new media, and personal tech.
about | subscribe | @ecbanks