Today, data centers are making big changes in how energy is used; let’s take a look at how far the industry has come.
Just seven short years ago, times were very different in our industry. Power grid demand experts feared we were heading straight into a capacity crisis. Whitepapers noted that data centers were becoming a significant electrical grid presence. Server deployments were happening at a maddening pace and were straining the power and cooling capacities of many of these specialized buildings, causing a data center construction boom that remained robust, even through The Great Recession.
As I.T. departments demanded more powerful CPUs, faster hard drives and faster networks, power consumption across all I.T. platforms was increasing significantly. Servers were routinely deployed for single applications and were sized for worst-case, momentary peak loads. This resulted in machines running at only 5-10% utilization 95% of the year, while power supplies for dual-corded servers were only running around 70% efficient.
Data center operators’ focus had always been availability, not energy efficiency. Any attempt at energy reduction was considered to be a risk. PUE was conceptual and just beginning to be understood, and no one tracked it in real-time. Fear of power grid strains was increasing. Did the U.S. and the world have the electrical capacity to provide power to these new types of constant-load buildings that our economies were becoming more and more dependent upon? In the U.S., even the EPA began attending our data center conferences and studying our industry. They studied our practices for a year and reported their findings to the Bush Administration.
The needle has moved significantly since then. What was once innovative has now become mainstream for new building designs.
In a 2008 article I wrote for The Uptime Institute, I mentioned energy saving opportunities in the whitespace, and how these had benefited the middle-aged Tier IV data center I managed for United Parcel Service.
You will find no surprises in the topics I covered in that article…
- Mechanical Systems for older data centers provide the best opportunities for energy efficiency improvements
- I stated that starting out on the raised floor (where the load resides) is where an energy reduction quest really begins
- Minimizing ‘bypass air’
- Using blanking plates in server racks to better separate hot and cold airstreams, which maximizes heat transfer through the servers and at the CRAHs
- This then leads to additional mechanical plant efficiencies
I also mentioned increasing the Chilled Water set point temperatures. This not only off-loads the chiller (saving energy), but also increases the number of annual hours on Free Cooling. (Is ANYONE still running a chilled water supply temp of 44F like what was designed in ’94? Hopefully not!)
Funny how in such a short time, those ‘innovations’ are now mainstream and represent only fundamental energy best practices. Today’s designs are accomplishing these objectives by default.
Check our blog next week for Part II where I will examine mechanical, electrical and I.T. improvements that are quickly becoming the new standards for data centers today.