Driving Power And Energy Management In Data Centers

Achieving efficiencies in data center management requires the distinction between power and energy.

By Andrew Sanchez

A few years ago when I interviewed for my first job in the DCIM (data center infrastructure management) world, the first question out of the hiring manager’s mouth was: “Can you tell me the difference between power and energy?” I tried to come up with something intelligent to say but soon realized I had no clue. Luckily for me the hiring manager was more interested in potential than actual knowledge.

Before we continue, it’s necessary to understand the difference between the two terms, power and energy.data center management

Power is the product of voltage and current. Voltage is the electromotive force (push) while current (measured in amperes or amps for short) is the actual number of electrons flowing through a conductor. Power is measured in watts. A watt is the current of one amp with one volt. Dizzy yet? You are not alone as these are not easy concepts to understand; not because they are advanced but rather because these electrical concepts are a century old.

Energy is power over time. It is the accumulation of fluctuating power normalized by hour so that it can be measured more objectively. As an example, take a 100 watt and a 40 watt lightbulb. If you turn on the 100 watt lightbulb for 20 hours you consumed 2 kilowatts per hour, or kWh [(100wx20h)/1000] which is the same energy as running the 40 watt lightbulb for 50 hours [(40wx50h)/1000]. Energy (kWh) helps normalize the consumption of different loads as in the two different size lightbulbs.

To make the concept of power and energy easier to understand, consider the following analogy.

Suppose you are driving down the freeway and keeping an eye on the dashboard. The speed is just like power as both are rates in real time. Energy is analogous to the miles travelled recorded by the odometer. Energy is measured in kilowatts per hour (kWh). Therefore, power is to speed as energy is to miles travelled.

Now let’s explore how these two concepts apply to data center management. In a typical data center, there are two operations. The first one is facility management, operated by electrical and mechanical experts who keep the whole place running. These experts maintain equipment like generators, air conditioners, transfer switches, and power distribution units with the goal of always supplying adequate electricity and cooling. The second operation is IT which is comprised of racks, servers and networking equipment which is maintained by network and hardware/software experts who keep the equipment humming along.

Let’s bring back the concepts that we discussed earlier: power and energy. The facilities side of the house is involved primarily in power management. The main role of facilities is to provide power and cooling so IT can keep the equipment running. Capacity in facilities equipment is rated in kilovolts-amperes (KVA). The facilities team keeps equipment well under the rated capacity and with redundancy in case of consumption spike. Reaching capacity in a PDU could mean a bunch of racks and the equipment within them powering off. Reaching capacity in a UPS could mean a bunch of PDUs going down and so on up the line. As you most likely know, facilities folks will tend to operate equipment in a way to always ensure remaining capacity.

There is just one little detail that makes a huge impact in the data center management operation: the electric bill is based on energy, not power. The utility meter keeps track of the kWh consumed during the month, and the utility company bills the datacenter accordingly. To make matters more complicated, the rate of cents per kWh may change depending on the time of day with the evenings being usually cheaper than the mornings. Some regions may have multiple energy prices per day depending on the total demand for energy on the grid.

To illustrate, assume your data center runs at an average of 2,000kW (2MW) at all times and that your energy cost is a conservative $0.09/kWh. That translates into a monthly energy bill of $129,600 assuming you have no spikes and your rates are always constant. The annual energy cost is $1,555,200 which may not sound like a lot compared to the total cost of the data center but it’s an operational cost so it counts against revenue and affects the bottom line directly. Nevertheless, the electric consumption, and thus cost, is not necessarily set in stone. There are usually ways to lower it. The only question is whose job is it to manage energy?

Let’s look at the teams that are already in place in data centers.

The facility management operation is there to keep things running comfortably well below capacity thresholds with redundant equipment to pick up the slack in case of sudden demand spikes. Facilities may be able to help reduce energy consumption with efficiency initiatives such as replacing older equipment with more efficient models or using the appropriate conductor for the amount of current. However, their goal is to keep things running; not lower energy costs. In other words, facilities are in the power management business which basically means keeping the data center running by managing capacity and load. It sounds simple but it’s a huge undertaking that requires juggling a lot of interconnected electrical equipment.

The IT side has done a lot to keep efficient operations and energy consumption constant. IT equipment manufacturers are continuously improving efficiency with every model which introduces that efficiency into the data center with every equipment refresh. Virtualization has greatly reduced the number of servers required in the typical data center operation which lowers the number of total power supplies required which in turn uses less energy overall. This is still not energy management. This is just a byproduct of improvements in technology.

Energy management is a business mindset, a strategy that looks at operations for ways to lower the energy consumption while still providing the same level of service. How? By identifying efficiencies and executing a plan that will lead to the improvements to the bottom line. Eliminating ghost servers, moving processing to locations with lower energy rates, adjusting temperature and humidity to keep equipment running; not to keep it cold and keeping transformers at high power factor levels are a start.

Whose job is it to manage energy? Ultimately, it’s everyone’s job.

Going back to the driving analogy, power management is like making sure the car doesn’t run out of gas or driving under the speed limit to avoid a traffic ticket. Energy management is like using the GPS to look for the shortest route to save gas and time while still arriving at your destination.

data center managementSanchez is senior systems engineer for Nlyte Software, a data center infrastructure management (DCIM) solution provider that works to seamlessly automate data center operations and infrastructure into an enterprise’s IT ecosystem. He works with data centers to help manage assets, energy, and infrastructure. Sanchez holds an MBA in IT Management from the Braniff Graduate School of Management at The University of Dallas.