Proper power capacity planning through DCIM

Proper power capacity planning through DCIM

Power capacity planning is becoming too complex for manual methods.

Much like hardware, physical space (footprint) and computer room air-conditioning (CRAC) units, electricity is a resource that is taxed according to the current needs of an organization. And just like every other resource in the data center, power has its capacity limits. Even a slight miscalculation in the supply and demand of electricity as new clients and equipment are added can result in shorts that induce costly downtime.

So how can data center managers accurately plan for power capacity – especially in a day and age when scalability and elasticity are more or less mandatory features? 

Where DCIM and power management converge

First, it's important to note that real-time power monitoring plays a critical role in solving the problem of planning for power capacity. This is because having precise measurements of power consumption throughout the entire facility will improve the quality of the data being used to calibrate for future loads. As TechTarget's Margaret Rouse put it, "the capacity planning process requires sophisticated load calculations both at normal and peak performance times."

With that said, analysis of these power metrics is what will ultimately provide the insights data center managers use to plan for power capacity. This analysis is far too sophisticated for manual calculations, which is where data center infrastructure management (DCIM) comes into the picture. The role DCIM plays in the data center is its ability to calculate massive quantities of data, and to interpret that analysis into easy-to-comprehend visualizations.

Beyond just providing real-time information regarding the current state of data center operations, this also enables data center managers to make forecasts. One example of such a forecast is power capacity planning – which is a critical convergence point for power management and DCIM. 

DCIM does the capacity planning math for you.DCIM does the capacity planning math for you.

From phase to rack: See it all with DCIM

Outages resulting from shorts can be widespread, or they can be localized to specific cabinets, depending largely upon how a facility is laid out. However, if power demand outstrips available electricity – even if the issue is isolated to a single power distribution unit – server downtime will invariably ensue.  

DCIM's power management capabilities solve this problem in two ways: 

  1. Real-time power consumption tracking informs data center managers how near they are to capacity at any given time of the day. 
  2. Predictive modeling allows for precision planning for future power loads.

That second item deserves a little more extrapolation. According to Data Center Dynamics contributor Matthew Larbey, high-performance computing (HPC) has become the more cost-efficient method for shouldering the strain placed on modern data centers. Of course, HPC requires higher-density power loads, since the computing power is significantly greater. Even then, managing power dense facilities saves money in the long run.   

Rather than building more data centers (which is an exorbitant undertaking) or leasing more space, organizations are finding that it makes more sense to manage these significant power loads. Part of achieving sustained success is being able to constantly plan for growth, which is where predictive modeling comes into the picture. Capacity planning determines at what point power load will reach or exceed limits based on current demands on infrastructure. Predictive modeling can complement this capability by forecasting the power that would be needed (both on a macro and micro level) to meet future demand under a certain set of parameters.

With so much focus placed on the benefits of scalability and flexibility, predictive modeling has become an extremely useful aid to capacity planning. Bundled together within a single DCIM solution, these key features let data center managers have a complete view of power consumption and load limitations – in the present, and in the future.