The Secret to Measuring and Optimizing Data Center Cooling Performance

Thursday, 15 July 2010 15:41

Written by Wesley Smith

How do you know if your data center cooling system is operating at its peak performance? Understanding key performance indicators arms data center managers with the power to assess, track and manage cooling system efficiency. Consider these steps to develop useful metrics for analyzing cooling system performance and create a baseline for optimizing operations.

Understanding your cooling specifications
Before any measurement or analysis can be performed, you must understand the cooling system specifications of your computer room. Without this baseline, no comparison exists when reading and analyzing environmental measurements.

Typically, cooling specifications are provided using the gross square footage of the computer room. This measurement, traditionally in watts per square foot, incorporates all of the floor space including the space not utilized by racks or other equipment.

Determine your wattage by cabinet
• Calculate the physical square footage that a cabinet consumes and multiply this by your cooling specification. For example, if your rack takes up 25 square feet (which includes space in front and behind the rack), and the cooling capacity of the room is 100 watts per square foot, then your cooling density per rack will be 2.5kW.

Understanding your IT load
Change is constant in the IT profession. Chances are the IT equipment that was used to create specifications for your data center no longer exists. During the life cycle of your data center, the IT equipment will change many times. Understanding the heat load generated by IT equipment is essential to cooling system performance.

Calculate heat load
• Review the manufacturer’s installation documentation for each server and add up the amount of BTUs per rack. This is the easiest method for determining your expected heat load.
• You can also estimate your heat load by using metered PDUs in your racks. To do this, take your amperage reading on your PDU and multiply this value by the voltage to obtain your watts per cabinet. Add up the wattage totals of your cabinets and this value can be used as an estimate of your heat load generated by your IT equipment. This information can also be used to help you balance the heat load across racks.

Create a Baseline using Environmental Tests
Once the engineering specifications are understood and the heat load has been calculated for the IT equipment, you can create a baseline for environmental variables in your room. Creating a baseline involves actively monitoring strategically placed temperature, humidity and air flow probes in the computer room over a given period of time. Not only will this provide a current snapshot of how your cooling system is operating, but it can also alert you potential issues before they become problems or incidents.

Measuring tools do not have to be expensive to be effective. For example, some tools monitor up to 16 temperature, humidity and air flow probes for under $1000. Pair this with an open source monitoring software package, and you can also trend against your computer room specifications and alert when points are out of range.

Measure the temperature and humidity of your computer room
• Place humidity and temperature probes in the cold aisles
• Place the probes in the middle of the aisle between the rows of racks
• Place probes, or utilize probes on your air handler units, to measure the temperature and humidity of the return air

These measurements will provide a good cooling profile of the computer room and allow you to compare the measured results to the computer room specifications. They can also indicate problems in the computer room before they impact operations.

Troubleshooting Data Center Cooling Performance
Data center managers must respond when computer room cooling specifications are out of operating range. The best defense against cooling system failures is to incorporate facility tests along side of scheduled cooling system preventative maintenance.
An easy test to incorporate is to profile the cold aisles of the computer room on either a monthly or quarterly basis. It should also be considered to perform this test upon a significant change to the computer room environment (for example, replacing a traditional server rack with blade servers).

To profile the cold aisles, place sensors in the cold aisles approximately every 4 to 5 racks apart and 5 feet off of the ground. Take a temperature, humidity and possibly air flow readings. Recording this data on the floor plan can indicate proper operation or trouble spots across the computer room floor.

If areas are found to be out of spec, this could be an indication of:
• Obstructions in the supply air path
• An imbalance of computer heat load
• Air recirculation where supply air is mixing with return air

Areas that are out of specification should be addressed either by self discovery or by hiring a consulting company to perform a cooling system analysis.

Understanding your cooling specifications, applied IT load and environmental variables will allow you to create a baseline that will yield a properly operating and efficient cooling system. Use these results to troubleshoot performance and find areas where improvements can drive efficiency.

About the Author:  Wesley Smith, Director of Operations, FIBERTOWN Tier IV Data Center

TrackBack URI for this entry

Comments (0)

Subscribe to this comment's feed