No matter how carefully a colocation provider designs and maintains the power infrastructure for a particular facility, the current operating model has created a weak link in the last mile.
The colo controls the power path from the utility to the rack, then turns over responsibility for power in the rack to the customer. Some customers choose to deploy ITE-grade rack PDUs (rPDUs) designed to reliably handle the high temperature condition that exist in many racks while others may choose lower cost and lower grade rPDUs that were never designed for such conditions.
In an ideal world, this wouldn’t be a problem for the provider. If a failure occurs in the rPDU, it should be no different than a failure in a server or switch: customer equipment, customer responsibility.
But is isn’t always that clear cut. The overcurrent protection in an rPDU generally doesn’t trigger as fast as the overcurrent protection in upstream breakers. That means when an rPDU fails, it can cause the breaker upstream to trip before its own overcurrent protection responds. As a result, the customer may place the responsibility for the failure on the provider.
The provider then has to spend time convincing the customer the original problem was in the rPDU. If they can’t do that, they may have to issue credits to keep the customer satisfied.
There are several solutions to this issue and now is the time to begin implementing them. While some providers may be concerned about disrupting the status quo during a time of growth, continuing to operate with an inherently “disconnected” power chain will only allow the current model to become more embedded and make it harder to change in the future.
One potential solution is for providers to own the critical power path through the last mile. Take the decision out of the customer’s hands and eliminate finger pointing in the event of a failure by adding the rPDU to the cost of the service. This also allows the provider to offer value-added services enabled by rPDUs, such as remote monitoring and outlet switching.
While some colos are already moving in this direction, it may be too big of a step for others. They may consider a more educational approach. Instead of taking ownership of the last mile, they can take a more active role in evaluating available options and educating their customers about the role of the rPDU in achieving the availability goals the provider and the customer share.
Many IT professionals haven’t taken the time to analyze differences between various rPDUs and how capabilities the rPDU enables, such as remote switching, can enhance the management of their off-site assets. They may choose a low-cost option to save a few dollars, not understanding those short-term savings could create long-term problems. But they are turning to colos for their infrastructure expertise and the rPDU, while now commonly seen as the responsibility of IT, clearly falls into the critical infrastructure category.
So, what does this education look like? It means applying the same due diligence and rigor to rPDUs as providers routinely apply to other critical infrastructure systems. Then, offering customers a list of qualified vendors whose manufacturing and testing processes, product range and capabilities, and certifications have been vetted by the colo for their ability to meet both the customer’s and provider’s requirements.
That process should include visiting the factory where select rPDUs are manufactured. This is a common practice for facility UPS and power distribution systems, but often overlooked for rPDUs.
Colo providers have been successful at delivering outstanding value and availability to customers today by developing deep expertise in data center infrastructure. It’s time to extend that expertise to the rPDU. In so doing, colos can strengthen their relationships with customers and enhance their ability to deliver on SLAs.