29 July 2012
Proper airflow and thermal management are two of the most critical aspects of designing and operating your data center, ensuring uptime of applications, consistent IT equipment performance and efficient operations.

In the past, data center managers would over supply a room with cold air to address small hot spots and perceived requirements for cooling the data center. Typical rack densities averaged 3 - 5 kW, thus the cooling requirements were moderate. Very few data center managers were pushing the envelope on operational efficiencies, they were simply smoothing over the imperfections.

Increased density

New technological advancements such as virtualization and cloud computing are increasing power rack density requirements, which are now reaching 15 - 20 kW. High density applications are reaching 25 - 30kW.

You are effectively driving the utilization of any server from 10% to 70 - 90%. This is taking the power density and airflow requirements and doubling or even tripling them. In the short term you have extra compute capacity, but now you are deploying servers that need higher power and more airflow.

Increasing densities make it less appealing to simply solve the problem by overcooling. The major problem with over cooling is the cost associated with it. Providing more cool air to the room requires the fans on air handlers to run faster, requiring more power.

There are many common sense best practices to deploy in the data center to improve thermal performance and airflow management. These include:

• The use of blanking panels in the racks

• Proper perforated tile placement

• Sealing up cable entry and exit points

• Separating the hot air from the cold air supply, which will help ensure the IT equipment is taking in the proper cold air, maximizing its performance and improving thermal efficiency.

Basics of thermal efficiency

The real issue for cooling servers and switches is delivering the right air temperature to the inlet for every piece of equipment and effectively removing the hot air. The first step is to implement industry standards and best practices at the room level and cabinet level.

Over the years, the approach to data center thermal management has been to over-cool a data center at considerable expense. As energy costs have risen, many data center professionals have raised the temperature to save costs, without knowing what effect this will have on thermal performance and operations.

To maximize the cooling capacity of your data center, you need to balance the deployment of IT equipment with the thermal management system's capacity to cool the equipment, ensuring that you maximize your data center's capacity while at the same time preventing downtime due to thermal issues.

Part of what makes this difficult is that the successful optimization of your data center requires an intimate understanding of something that you cannot see - Airflow!

The correct amount of cold air needs to be delivered to the air intakes of the IT equipment and the hot exhaust air needs to be returned into the cooling cycle efficiently.

Thermal strategy

To cool your data centre effectively, you need to develop a cooling strategy based on detailed knowledge of their data center's thermal dynamics.

Therefore to maximize the efficiency of the system you must first uncover the root causes of the cooling problems plaguing the data center - remember "hot spots" are a symptom, not usually the primary cause of the problem. Without fully determining this, it can be difficult to choose the appropriate course of action, in order to fully resolve the cooling issues, let alone improve efficiency.

A key consideration should be the use of Computational Fluid Dynamics (CFD) analysis, in order to identify the root cause and then look at the solutions and cooling architecture that will optimize your data center for maximum performance and efficiency.

How can this service deliver a return on investment?

  • Helps you to choose the most optimal long-term solution to accommodate the impact of future IT equipment deployments
  • Provides a baseline of your current thermal management issues and provides solutions based on "what if" analysis
  • Determines the most optimum solutions based on a proven analysis methodology instead of trial and error
  • Identifies optimum strategies, specific to your data center - not the cooling vendor
  • Identifies the optimum set point temperature for your data center, based on ASHRAE thermal guidelines
  • Unlocks cooling capacity, by balancing the airflow

Once you have a data center that is optimised, the next step would be to ensure you had an efficient Data Center Infrastructure Management (DCIM) tool in place, to fully monitor and report, real time, on the environmental conditions within the facility.

David Hughes is director of professional services, EMEA, at Panduit. He has more than 25 years experience in designing, deploying and managing IT infrastructure solutions, working across most vertical markets and industry sectors.

© Zawya 2012