Power monitoring: If it's not 99 percent accurate, then it's not for you

Power monitoring: If it's not 99 percent accurate, then it's not for you

A 5 percent margin of error won't cut it in the data center.

Up until fairly recently, a 5 percent margin of error for power metering might have been acceptable. However, higher density data centers mean more significant power loads, and that means even minute discrepancies in power consumption readings can have far-reaching negative consequences. Anything more significant that a 1 percent margin for error can eat into your bottom line, result in unbalanced loads, or even skew numbers when planning for future capacities. 

Here's what's at stake:

Billing accuracy

When utilities bill clients, they do it with 99 percent accuracy because their entire business model depends on getting paid for the product it delivers.

"Be done with sneaker reports once and for all."

Likewise, data centers – especially colocation providers – shouldn't be eating the costs of powering their tenants' equipment. The goal is to turn a profit, and that's not easy to do if you can't be 100 percent (or at least 99 percent) certain that the asking price is accurate. 

In other words, incremental power metering is a loss waiting to happen. The only way to ensure 99 percent billing accuracy is by continuously monitoring power utilization in real time. Be done with sneaker reports once and for all. 

Load balancing

Flexible resource allocation is crucial to handling peak usage times in any data center. However, higher rack kW in high-density facilities means power load distribution must be, or at least should be, micro-managed. Continuous power availability needs to be maintained to prevent shorts, but also controlled to avoid unnecessary utilization of energy that could inflate your power usage effectiveness (PUE) rating.  

Ideally, then, power monitoring should be as precise as it is continuous. By aggregating real-time, outlet-level power utilization metrics from each power distribution unit (PDU) in your facility, you can:

  1. Gain a bird's-eye view of how power moves through your facility and when utilization spikes. 
  2. Make precise, granular decisions that allow you to simultaneously optimize performance, reliability and efficiency.  

This will help facilitate the IT elasticity that the majority of today's businesses have come to expect.  

Accurate metrics precede accurate decisions.Accurate metrics precede accurate decisions.

Capacity planning

Scaling the data center is historically one of the most complicated endeavors. It isn't just about calculating the necessary IT load. It's also about ensuring that you have the footprint, the cooling capacity, the power distribution infrastructure, the electrical redundancy and the proper environmental and airflow monitoring needed to support the introduction of additional equipment. 

Increasingly, businesses are relying on data center infrastructure management (DCIM) software to do some of the math for them. But for this to work, they need to feed the machine with data, and a lot of it.

Real-time, 24/7 power and environmental monitoring provides that data, and not just in incremental bits and pieces that help you connect the dots. Specific correlations such as the impact that additional servers might have on rack temperature can be easily depicted via an intuitive graphical interface. The production of these reports can then directly influence the facilitation of smarter capacity-management decisions that will optimize complex facility operations as new variables are introduced to the equation.  

If you're a facility manager, we don't need to tell you this, but that's kind of a big deal. And you shouldn't settle for anything less.