How to Rate the PUE and DCIE of your Datacenter

Power Usage
Power Usage

Power usage efficiency (PUE) is the total power your data center consumes over the energy your computer equipment uses. Data center infrastructure efficiency (DCiE) is the reciprocal value expressed as a percentage. 

The Uptime Institute has said the average data center has a PUE of 2.5. The DCiE of that would be 40% which amounts to overhead power consumption. These two indicators are therefore valuable benchmarks for determining the energy efficiency of an equipment room.

Let’s Take a Step Back and Put Power Usage Effectiveness and DCiE in Context

We can’t do without our IT equipment and still be able to deliver our services. We also need a secure environment to protect our data, while we complete business transactions, do our R&D and keep our company growing. However, data is a growing cost factor we have to manage.

Moreover, the bigger we get the more important it becomes to protect our business information. So we end up with an impressive collection of computer servers, hubs, routers, wiring patch panels, and other network equipment consuming electricity at a cost.

Our data center may have started out in life as little more than a server rack in a closet using minimal power. However, since then it may have grown to become a computer room, a server room, or even a PUE data center.

Our closet in the corner now has lights, climate control to regulate temperature and humidity, security, backup power, and access control. Therefore, the infrastructure may be consuming more electricity than the IT equipment it exists to serve.

How PUE and DCiE Help Us Keep an Eye on Data Center Costs

It’s a relatively simple matter to install separate electricity meters for an IT facility, and for the data processing equipment inside. A quick monthly check enables us to see how our energy ratios are trending, and whether our efforts to manage them are bearing fruit. 

Data processing converts some of the electricity it receives into heat. At the same time, it does not function optimally in a hot environment. This phenomenon drives the cost of cooling which can become a significant outlay.

The processing equipment is therefore the right place to start when managing these costs down. Our Power Usage Effectiveness and DCiE scores can tell us how urgently we need to do so. Let’s find out how to establish what our actual ratings are.

How to Determine Your DCiE Rating

Your score may not be entirely representative if you are doing this for the first time. That’s because your electricity consumption varies depending on where you are in your business cycle, and what the ambient temperature is outside. 

We recommend taking monthly readings, trending them on a spreadsheet, and sharing them with your people. They are probably better equipped to understand the operational reasons behind the peaks and valleys.

  • Record the energy reading as close as possible to where the supply enters the data center, noting any exceptional energy events during the period. By way of example, there may have been a power outage, or maintenance work done in the facility.
  • Record the energy reading as close as possible to where the supply enters the equipment racking, but definitely after it has passed through power conversion, switching, and conditioning. As before, note any unusual energy-related events during the review period.

Determine the differences from the previous readings and take any action indicated by what you discover. Ideally, you want to see your equipment consumption trend down with better cooling strategies, and your overall DCiE score improving.

Your equipment energy score is absolute because no two computer rooms are the same. However, your overall factor becomes relative when you compare it to these norms provided by 42U.

Why You Should Delegate This Task to Your Administration Manager

There are various, sometimes opposing roles in a large data center. The core activity is the data itself while power backup and cooling equipment often take up the most time. The various teams, therefore, have different priorities and goals.

Hence you probably have a team focusing on managing the data center, especially the backup power and cooling. While a second one may see the IT equipment deserving priority.  

The facility manager usually rules on matters of infrastructure environmental issues including power, cooling and airflow, while IT people are often primarily interested in the health of the servers and networks.

An independent person – such as the admin manager – may therefore be better positioned to manage the overall energy consumption, including the data center and the equipment it serves. This is only a suggestion for your circumstances may well be different.

Why You May Get a Frown When You Mention This

Power Usage Effectiveness (PUE) and its reciprocal Data Center Infrastructure Efficiency (DCiE) enjoy support from opinion-makers such as Green Grid and Uptime Institute. However, some creative IT people believe they over-simplify things according to Storage Servers.

Nonetheless, both are useful indicators of how our energy consumption is trending, and how it compares with our peers. However, the ratios do not factor in, and factor out the unique differences compared to ‘average data centers’ (if there is such a thing).

Power Usage Effectiveness and DCiE Takeaways

DCiE and PUE are management information, not cast in stone. It’s what you do with the information that matters after you compare your data center consumption with your infotech load. Are you satisfied with what you see in the context of the greater reality you juggle as a manager?

At the very least, you should include PUE or DCiE in the key performance areas of your data center operation. That’s because they provide you with a vehicle to compare the present and the past with the direction you are taking your organization.

Your power consumption has a bearing on your energy back up and cooling strategies. Hence there could be a multiplier effect unless your people know you are monitoring what happens in the server room.