CALL FOR INFORMATION +44 (0) 1509 261100

Power and Cooling Capacity Management for Data Centres

It is absolutely essential to measure and correctly predict power and cooling capability at the rack level of any high density data centre environment to avoid unexpected challenges such as overheating, overloaded infrastructure or loss of resilience and redundancy. Without this measurement it is impossible to ensure a predicable performance optimised for your data centre.  
Gartner Inc's latest research shows that the majority of network managers are unsure of the exact loading, power and cooling capabilities of their data centre. It’s the installation of equipment beyond the limits emposed by existing infrastructure that causes most downtime. According to Gartner, most data centre managers are unaware of the loading and current power and cooling capability of their data centres, even at a total bulk level. Installing equipment that exceeds the design density of the data centre, and the resultant stresses on the power and cooling systems, are causing downtime from overloads, overheating, and loss of redundancy.
Capacity Management
This process ensures that equipment is installed at the right time, in the right volume, and at the right price and that it is deployed in the most efficient method. This depends on:   

  • Provision of accurate capacity predictions
  • Provision of sufficient but not excessive capacity to meet business needs

Detail is often sought from many areas in order to establish exactly what IT systems are (and will be) required, what level of power and cooling is demanded and what level of resilience will be required. Cost is also key. on365 offer a capacity management capability specifically aimed at determining power, cooling and space requirements for data centres of all sizes.  

This starts with some simple questions to determine what’s required by the business:  

  • Where should incremental devices be situated to avoid impact on existing equipment and resource,
  • What would be the most efficient location, from a power and cooling perspective, to place,
  • Will installation threaten existing redundancy or back-up protocols,
  • Will contingency still be available during essential maintenance,
  • How will heavy load devices such as high capacity servers increase the demand on power and cooling infrastructure,
  • What happens when available power/cooling supply is exceeded and when will this occur?

There are three main factors driving demand for proactive management of this type, these factors are: 

  1. Increasingly high capacity equipment deployed in increasingly high density environments
  2. Deployment decisions based on TCO (total cost of ownership)
  3. Business/User demand for new technology.

Capacity Supply and Demand
The first step to proactively managing capacity is the ability to quantify supply and likely demand of power and cooling. A big mistake is to take information provided at room level to determine these figures however resulting datsa is far too broad to be accurate and when taken from manufacturers information relating to specific products the data assembled is too specific. The ideal method is to establish power and cooling demands at the rack level.

on365 approach cooling and power demand calculation using 4 distinct calculations:  

  • As-configured maximum potential demand
  • Current actual demand
  • As-configured potential supply
  • Current actual supply

System-level Capacities
Existing power and cooling devices are unlikely to be aligned with specific racks, more likely associated with blocks of racks or the entire room itself. With this in mind it becomes increasingly important to not only understand the specific racks demand and capability (as above) but also the aggregate demand placed on existing power/cooling infrastructure by multiple racks. A status of excess supply is essential at rack or room level to ensure no failure and down time, excess supply can be achieved in four specific ways:  

  • Spare capacity
  • Idle capacity
  • Safety margin capacity
  • Stranded capacity

Functions for Managing Capacity

on365 determine a number of key measurable functions within the management process, these functions are as follow: 

  • Monitoring of performance
  • Monitoring of workload 
  • Predicting supply
  • Predicting demand 
  • Accurate modeling

Deployment of capacity management software across the network is the only feasible way to tackle the increasing complexity and demands from data centre power supply. When successfully deployed, networked instrumentation and management delivers the following benefits: 

  • Simple and graphical presentation of capacity data
  • Simple translation of results into a capacity management plan
  • Alerts on violations and faults in excess of this plan
  • Facilitates modeling against future requirements.

When to start the capacity plan
This should be done at the design phase of the data centre installation. Using the methods above, on365 develop a plan that works in today's scalable data environment to ensure sufficient and approporiate supply of power and cooling. Oversized capacity can be equally damaging to the predictability and cost of a data centre environment.

On-going monitoring of Power and Cooling for IT Devices

Careful planning and insistence on sufficient power at the rack and room level means that on-going maintenance becomes less important. Sometime it’s beneficial however to maintain a basic level of realtime management of devices, the reasons for this could be: 

  • Early warning on unusual power consumption characteristics at device level
  • Qualification of minimum and maximum flow rate and power characteristics at device level
  • Alerts for unusual airflow, or unexplained fan operating modes
  • U-space utilisation and capacity within the rack level 
  • Measure impact of MAC’s (moves, adds and changes)

on365 realise that not all data centres have specific resource or management capability for deployment of this kind of in-depth planning. We therefore rationalise the process at all levels to suit the level of information/resource available yet aim to provide a scalable plan “structure” that can grow with the business and it’s IT requirments.

It is clear that capacity management is an essential part of planning and execution in all sizes of data centre However the complexity of plan, and dependence on the plan grows in parallel with density. If capacity management is deployed at the initial installation phase and developed in sync with future deployments then it will continue to provide critical capability date not easily determined with traditional monitoring methods.  


  • Pelco
  • Schneider Electric Partner
  • APC Partner