Installing equipment in a data center without considering its loading and current power as well as its cooling capacity can put stress on an IT system and result in downtime from overloads, overheating and loss of redundancy.
But in most cases, according to Gartner Inc., data center operators don’t have the information they need to strategically add equipment to new or existing data centers.
Capacity management – defined by the IT Infrastructure Library (ITIL) as the discipline that ensures infrastructure is provided at the right time in the right volume at the right price, used in the most efficient manner – is vital to ensuring predictable performance and optimizing the use of the physical infrastructure.
The key is to determine in the system planning process what IT systems are or will be required by the business, what power and cooling infrastructure will be required to support those systems, what level of contingency will be needed, and what the cost of this infrastructure will be.
Critical success factors include providing accurate capacity forecasts as well as providing appropriate capacity to meet business needs. Capacity management data can be used to reduce total cost of ownership (TCO) and increase efficiency.
In many cases, data center operators don’t have the information they need and install equipment that exceeds the density design of the data center. Stress on the system is caused by:
- High-density equipment: The vast majority of data centers are designed for a power density of less than 2 kW per rack. Yet, fully populated racks can draw between 6 kW and 35 kW per rack (more than 10 kW per rack can be considered high density). The result is downtime cause by overloads, overheating and a loss of redundancy.
- Total cost of ownership (TCO): Estimates indicate most data centers could hold up to 30 percent more equipment if power and cooling capacity was more efficiently managed. Inability to fully utilize this capacity reduces efficiency and drives up power consumption by 20 percent or more compared to a properly managed system. TCO pressure reduces efficiency, causes unseen waste and contributes to unused capacity.
- Rapid pace of change: A typical data center experiences an equipment refresh cycle of less than three years. In any data center, equipment is added or removed on a daily basis. In addition, the power requirements of system devices can vary minute-to-minute as the data load changes.
Each of these factors calls for the need for information to operate a data center in a more predictable manner. By providing data center operators with better information, they can more effectively lay out equipment to avoid these stresses.
But how can the necessary information be gathered?
Capacity management allows the user to determine the supply and demand for both power and cooling. A systematic approach to capacity management will generate information necessary for more effective operation of a data center.
Supply and demand
Power and cooling supply and demand information can be gathered at one of three levels: the room level, the rack level or the device level. At the room level, the information is too broad and doesn’t provide the level of detail required by capacity management. At the device level, the information is too detailed.
By determining power and cooling supply and demand information at the rack level, needs can be more effectively measured and budgeted. Quantifying this information takes place in four different ways:
- As-configured maximum potential demand
- Current actual demand
- As-configured potential supply
- Current actual supply
For more information on an effective model for managing capacity, see APC by Schneider Electric’s white paper, Power and Cooling Capacity Management for Data Centers.