All IT teams we work with today are struggling to understand ways in which they can increase the energy efficiency of their data center operations. Unfortunately, the scramble to demonstrate some quantifiable decrease in energy in the data center has led to uninformed decision making, impetuous investments, and increased frustration on the part of IT teams in determining the most effective means of creating an energy efficient operation.
These frustrations are certainly warranted. As energy efficiency in the data center has become a ubiquitous industry goal, the number and complexity of solutions available to increase efficiency have risen exponentially. Determining the best solution for a given IT environment is challenging, especially as case studies continue to emerge detailing innovative solutions—outside air as a primary cooling infrastructure, nearby rivers as a heat rejection mechanism—and much more.
In our first meetings with most IT teams, they want to discuss these types of complex solutions and ways in which they can leverage these dramatic changes in their data center. Their willingness to think unconventionally is commendable and an important part of the process. However, in doing so they are also losing sight of the information that is most critical, most useful, in determining how energy efficiency can be achieved: understanding what they have today.
Our message to IT teams is simple: get to know your equipment. If you want to make informed decisions about energy efficiency in the data center, it’s not enough to know the country the server is named after and what its IP address is. Y
ou need an inventory of each individual piece of equipment and, more importantly what its run rate load is. In watts. Most likely, you started off keeping this detail when you first built your data center 7 years ago. Then there were those 25 Windows servers your team had to work overtime to burn in to support a mission critical engineering project in 2006. Half of which were then decommissioned as part of your virtualization project in 2008. All of which, you realize, are still racked, turned on, and drawing power in your data center.
The first step in making impactful reductions in energy consumption in your data center is to fully understand your environment today. Create a load profile for your data center, segregated by rack, that documents the server name, make/model, and run rate load. This load information differs greatly from the “nameplate rating” you’ll find on a spec sheet. Leverage your relationships with equipment manufacturers or use online tools (HP, Dell, and APC and many more have these resources available online) to gather what the actual consumption of the server is under normal operating circumstances. Here is an example:
| Rack Number | Device Name | Make/Model | Rated input voltage | Rated input current | Load (watts) |
| 1 |
KVM01 |
Avocent DSR2020 |
100 to 240 VAC |
|
27 |
| |
VUS-GLLMS-02 |
HP Proliant DL380 G4 |
100 - 132 VAC, 200 -240 VAC |
7.5 A (100 VAC), 3.8 A (200 VAC) |
493 |
| |
VUS-GLLMS-03 |
HP Proliant DL380 G4 |
100 - 132 VAC, 200 -240 VAC |
7.5 A (100 VAC), 3.8 A (200 VAC) |
493 |
| |
VUS-GLSEC-01 |
HP Proliant DL380 G5 |
100 to 132 VAC, 200 to 240 VAC |
10 A (100 VAC), 4.9 A (200 VAC) |
644 |
| |
VUS-GLUSB-03 |
HP Proliant DL380 G5 |
100 to 132 VAC, 200 to 240 VAC |
10 A (100 VAC), 4.9 A (200 VAC) |
644 |
(Note, table does not contain actual data and is for illustrative purposes only)
The product of this effort is an empowering piece of information for making decisions about efficiency in the data center. For example, one of our customers recognized an extreme imbalance in how the kW loads were spread across the racks in their data center. More importantly, they had been struggling to control hot spots around what their load profile showed as higher density racks and had installed a ceiling-mount cooling unit to try to neutralize the heat. After review, they identified other racks in the data center where systems could be moved to, creating a better balance in load across all racks. This relocation enabled them to turn off the ceiling mount unit.
Making significant improvements to energy efficiency in the data center is something that undoubtedly requires a comprehensive strategy, outside-the-box thinking, and an understanding that implementing these changes requires an ongoing commitment. The best plans are pragmatic ones outlining a roadmap for small changes that collectively help achieve your efficiency goals. More often than not, you can make considerable strides through small changes in your existing environment that require minimal investment. The only way to identify these changes is to understand exactly what you have today. So, if you care about energy efficiency in your data center, do yourself a favor and get to know your IT equipment again.