4 rack PDU functions that improve data center uptime

4 rack PDU functions that improve data center uptime

Keeping the lights on in the data center is half the battle.

Data center downtime has never been costlier in our hyper-connected world. On top of the $8,900 per minute lost as a result of business disruptions, organizations must factor in remediation costs and the reputational damage that follows. For example, the parent company of a major U.K. airliner saw a 2.8 percent stock price drop after a power system failed in one of the company's critical data centers this May. 

With so much at stake, data center managers are under immense pressure to all but guarantee business uptime. One of the best starting points in this endeavor is power distribution infrastructure, and specifically power distribution units (PDUs). Next-generation PDUs supply many more functions than the basic power strips of yesteryear. Here are four that have great potential to improve data center uptime: 

1. Color-coded A/B feeds 

Architecting redundant power infrastructure is fundamental to making sure the failover to a secondary power source is seamless. These A/B feeds will be supported by two PDUs per rack or cabinet (depending on density). In a high-density setup, the quantity of PDUs really adds up, making power load balancing between supply feeds more delicate. All it takes is one misplaced PDU to cause a short capable of inducing unplanned downtime.    

The best fix for this problem is to color code PDUs to ensure clear, quick and easy feed identification at all times. Getting your wires crossed in the data center is a silly – but costly – mistake with a simple solution: color-coded PDUs. 

More power cables mean more PDUs, and that means greater risk for supply power feed mishaps.More power cables mean more PDUs, and that means greater risk for supply power feed mishaps.

2. RSTP

Speaking of PDUs adding up, data center managers have found daisy-chaining extremely useful for the number of connections between PDUs and network switches. These Ethernet-based connections are critical for monitoring power usage in real time. If they fail, or if the secondary power infrastructure isn't wired correctly, you risk losing remote visibility into the performance of your power infrastructure.

Cue rapid spanning tree protocol (RSTP). The beauty of RSTP is that it allows for two connections to a set of daisy-chained PDUs, once at the beginning, and again at the end. Traditionally, this was impossible, as it would result in bridge loops that would flood the connection. However, RSTP was developed specifically to enable a redundant network connection without causing these loops. Thus, if one connection to the switch fails for any reason, the other stays active, ensuring that you never operate in the dark. 

3. Up-to-the-second alert systems

"A stitch in time saves nine ... thousand dollars."

If electricity is the lifeblood of the data center, PDUs are the arteries. Data center managers must therefore be alerted the second any component of that electrical circulatory system is jeopardized – whether that's caused by an anomalous voltage spike, unbalanced power load, a short or something else (this is yet another reason why RSTP is clutch).

Remote power monitoring through the PDU is the most effective way to keep data center staff in the loop. By aggregating power usage and comparing that against allowable thresholds in real time, intelligent PDUs provide yet another feature, which is instantaneous, remote alerting. Designated staff automatically receive SNMP, SMS or email-based alerts the second something goes wrong. It's like they always say: A stitch in time saves nine ... thousand dollars.  

4. Environmental monitoring through PDU

Power-related failures tend to be the more notorious causes of data center downtime, but they don't stand alone. Overheating servers, humidity, condensation, flooding and fires have all been known to precipitate critical equipment failures. The secret to staving off these disruptions has been to monitor environmental conditions at strategic positions on racks.

An equally effective alternative that reduces footprint is to use intelligent PDUs with ports for hot-swappable intelligence. In addition to monitoring power usage, data center operators can achieve robust environmental monitoring at the rack PDU level. The same aforementioned benefits of power monitoring – redundancy, automatic alerting – help ensure that safe environmental thresholds are maintained. And while even these capabilities can't guarantee downtime will never happen, it certainly spins the odds in your favor.