Choose quality PDUs to improve data center uptime

Choose quality PDUs to improve data center uptime

Durability, reliability and functionality make for a strong PDU.

A data center is like a finely tuned motor. For it to run smoothly, every individual component must be perfectly calibrated and designed to be able to fail without disrupting critical functionality. 

But much like a world-class trapeze artist doesn't use a safety net as an excuse for sloppiness, redundancy is not the same as reliability. Data center managers must understand that having A and B power feeds is not enough. Reliability must be embedded into the very circuits of their power infrastructure, and especially in their power distribution units (PDUs). 

The question, though, is how does a data center manager vet for PDU quality in such a saturated market? More specifically, how do they make sure that the PDUs they select will improve data center uptime?

Trusted build quality

By trusted build quality, we mean a lot more than just a sufficient ISO score. Proper conformance marks are certainly important, but they're not distinguishing features. 

There are three key indicators that a PDU will do what it was designed to, and for the entire length of its warranty: 

  1. Highly accelerated life tests: More commonly referred to as HALT testing, this refers to the acceleration of stress conditions on a piece of hardware. Through HALT testing, manufacturers can identify predictable failures. This helps them verify the quality and reliability of a PDU's build, and also make projections about its lifespan. 
  2. UL certification: The purpose of a UL mark is to show that an independent third party has tested and validated claims made by the manufacturer. A UL mark also assures that the product in question conforms to safety standards. 
  3. Factory testing: After a product design that has been thoroughly tested is built, it should also go through factory tests before shipping to the client. This guarantees that a particular unit will do everything that it was built to do. As such, they may be scenario-based – for instance, does a monitored PDU accurately record current in real time?

Then there is the question of the warranty itself. Industry-wide, many PDU manufacturers will offer a warranty of two years for a basic PDU, and that simply isn't good enough. These power strips should last you a lifetime, and if a PDU manufacturer says otherwise, you may want to reconsider doing business with them.

When it comes to intelligent PDUs, that warranty should ideally be between three and five years. Again, there simply is no excuse for why a manufacturer would only cover their most basic products for two years when best-in-class manufacturers are covering their state-of-the-art PDUs for as long as five years in some cases. 

Long-lasting, reliable PDUs are crucial to creating dependable power infrastructure. Long-lasting, reliable PDUs are crucial to creating dependable power infrastructure.

Remote switching

Speaking of state-of-the-art PDUs, data center intelligence goes hand-in-hand with data center reliability. Basic PDUs still have their place in the data center, but ideally, operators should have greater visibility into the performance of mission-critical equipment. For colocation facilities, real-time, remote power monitoring is essential for accurate client billing. For data centers in general, the ability to track energy consumption patterns as in-depth as an individual server or switch adds an entirely new layer of quality: visibility into the performance of your facility both on the macro and micro levels. 

"Switched PDUs take quality to the next level."

Switched PDUs take that quality to the next level. Individual outlets on a specific PDU can actually be disabled remotely if, for instance, a particular server is experiencing a problem and needs to be power cycled. Consider another example, in which the monitored current is coming dangerously close to exceeding a pre-designated threshold for a certain rack. With a switched PDU, technicians will automatically be notified of this issue. They will then have the option of dealing with the situation remotely, perhaps by cutting power delivery to non-essential servers in that row until they're able to find a more permanent solution. In a day and age when a single minute of unplanned downtime can cost nearly $8,000, this level of remote visibility and control can prove a godsend.  

The secret to optimizing power infrastructure is to focus first on the reliability of your equipment. Has it been thoroughly vetted by the manufacturer and validated by third-party inspectors? Is the vendor standing by its work with a competitive warranty?

Then ask the next-most important question: How has this PDU built on that base of reliability? If the answer to both questions is to your liking, then congratulations! You've found your new vendor