Introduction

The main reasons for implementing a green data center strategy are to reduce costs, increase efficiency and to reduce the impact operating a data center has on the environment. IT organizations continue to be under increased pressure to reduce the total cost of ownership of running a data center. The constant challenge is to reduce annual operating expenses and limit capital expenditures. Cost, efficiency and environmental objectives can coexist and be mutually supportive for the IT organization. Increasing the efficiency of the IT infrastructure in the data center delays or eliminates the need to add more servers, storage and infrastructure such as additional power and cooling. Designing and operating a green data center improves the sustainability by limiting the impact to the environment when building the data center and improving the energy efficiency helps to reduce the carbon footprint and reduce costs. While the goals for implementing a green data center are good for any corporation, metrics to measure the success of your green strategy need to be identified. Processes also need to be created or modified to implement the metrics.

Data Center Power

The mission of the data center is to provide a highly reliable, available, secure and sustainable facility to house platforms and networks that enable IT organizations to meet the corporate computing and communication business needs. Data centers are usually the largest consumer of energy for the entire organization. The total power for the data center is the amount of power delivered to the data center by the local power utility company over a period of time. Most data centers that are mission critical in nature utilize dual power feeds from separate utility sub-stations. Power is defined by the formula (Volts X Amps) = Power. Power used over time is measured in kilowatt-hours (kWh) or 1,000 watts used for one hour. Figure 1 is an example of a data center with dual power feeds:

(diagram)

The components within the data center supporting the mission critical IT infrastructure for stable power and cooling for the raised floor typically consists of the following types of equipment:

• Cooling
  • Chillers
  • Pumps
  • Computer Room Air Conditioning units
  • Cooling towers
  • Cooling units
  • Chilled water storage tanks
• Power
  • Uninterruptible Power System (UPS)
  • Switching gear
  • Power Distribution Units (PDU)
  • Generators
  • Battery Backup

The following diagram depicts the data center infrastructure equipment:

(diagram)

The demarcation of the infrastructure equipment and the IT equipment is often the Power Distribution Unit (PDU). Organizationally, the building facilities group is responsible for operating and maintaining the data center infrastructure equipment. IT is responsible for all the platforms on the raised floor that connect to the PDUs. The data center IT platforms consist of the following types of equipment:

• Servers
  • UNIX
  • Web
  • Windows
  • Linux
  • AIX
• Virtual hosts
• Business intelligence platform
• Storage area networks
• Network area storage
• Data network elements
• Voice network element
• Virtual tape libraries
• Virtual tape servers
• Robotic tape libraries

The total power used in the data center is the sum of the power for the infrastructure equipment and all the IT platforms.

Where does the power go?

In a report to Congress on Server and Data Center Efficiency in August, 2007 the Environmental Protection Agency (EPA) estimated that “The power and cooling infrastructure that supports IT equipment in data centers also use significant energy, accounting for 50 percent of the total (power) consumption of the data center.” The EPA estimated the power consumption of the IT equipment using a bottom-up approach based on publicly available data for servers and data center equipment. A much easier approach, rather than estimating power consumption for all data center equipment, is to use the following formula:

(Total data center power) – (power to the raised floor) = (power for power and cooling infrastructure)

Total data center power is the total kWh based on the bill from the power company. Power to the raised floor can be measured from the switch gear. A simple example is 1000 kWh – 400 kWh = 600 kWh. In this example, the IT equipment is using 40 percent (400/1000) of the total data center power. From this example, the distribution of power for the facilities equipment and the IT equipment might be as follows:

(diagram)

Creating an IT platform inventory

From the switch gear and UPS we can determine the total power usage for the IT platforms. But how do we determine what platforms use how much power? How does the platforms power consumption match our business needs? To determine how the IT platforms use power on the raised floor, we need to conduct a 100 percent physical inventory. The asset management process for financial management is a good starting point for developing a list of all the IT assets in the data center. From the asset management data, conduct a physical inventory of all the platforms connected to the UPS. The next step is to supplement the asset management data with location and power specific data for each asset. Examples of data elements to add to the asset data are as follows:

• Location data
  • Grid location on the raised floor
  • Rack number
• Power data
  • PDU number
  • PDU panel and circuit
  • Power strip information
  • Measured KW (actual)
  • Plate KW (rated)

Measuring Power

Measuring the power for the IT equipment is combination of art and science. The art is using platform plate data to estimate power and the science is to use a meter to measure the actual power draw. Most data centers have hundreds of platforms and measuring all those devices on a periodic basis is not always feasible. Using the plate data for low power draw devices is a reasonable approach and provides a good estimate for the inventory. Plate power data is the power draw specified by the manufacturer as the maximum KWs for a platform. Using a watt meter you can measure the actual power draw for a platform or you can have your data center power contractor measure the major platforms for the actual power draw. Plate data estimates are good for small platform like stand-alone servers. For larger platforms, actual power measurements during prime-time usage periods are needed to determine the power for large IT platforms like a storage area network. Figure ## is an example of where to measure the power for a single connection platform:

(diagram)

Large platforms will have multiple power feeds so the total power will be the aggregate of all the measurements. After we have measured the power and entered the data into our inventory data store, we are ready to establish our baseline and develop our plan for reducing our power consumption.

Developing the action plan

After all the platforms have been inventoried, the next steps are as follows:

1. Define the baseline metric for the total power used by the IT equipment in the data center
2. Identify the top ten platforms by power usage
3. Identify total power usage by platform

The key metric for implementing a green data center strategy is the total power in kilowatt hours for the IT equipment on the raised floor, or stated another way, all the IT computing and network platforms connected to the UPS known as your critical load. The total KWH metric is easily measurable and can be assigned to the data center infrastructure group for accountability.

The metric becomes the indicator for the success of reducing the power consumption,, cutting the utility bill and improving the environment. The total KWH metric is easily measurable, understandable and actionable within IT.