Improving Architecture for High Density Data Centres

The environmental implications of wasted power through inefficient Data Centre power and cooling strategies are huge. Globally the power wastage from unused IT infrastructure is likely to be greater than 50,000,000 megawatt hours and is becoming an every greater financial and environmental burden on industry. on365 can help deploy Data Centre architecture that actively cuts down on wastage from power and cooling infrastructure, and in-doing so greatly increases the electrical efficiency over the entire installation.  

Focusing on the IT equipment is traditionally only half the story. Only half of a Data Centre power consumption goes to IT loads and devices, the other half is consumed by the support infrastructure in and around the Data Centre, this could be the lighting, cooling, security and electrical power systems. on365 also have a strategy for optimising the power consumed by support infrastructure.  

Computer equipment manufacturers are pioneering new solutions, such as virtualization, that will dramatically reduce the amount of IT equipment required to perform specific functions. This will bring a significant reduction in IT-load power consumption.  At the same time however, the trend for higher density Data Centre and time varying power draw is having a negative impact on tose same Data Centre power and cooling systems

How much does a higher density or varying IT load reduce data centre efficiency?
 
A well designed Data Centre can offer a tangible opportunity to increase power and cooling efficiency, even in high density environments, as long as a “smart” row based architecture is deployed.

on365 have the capability to reduce the consumption of Data Centre cooling and power resources by up to 50%  
 
We achieve this by focusing on the most effective elements of Data Centre design:

• Engineering design of individual devices
• Power distribution
• Inter-component communication and coordination
• Cooling strategy
• System planning
• Management tools

on365 installations that combine these elements as an integrated system show dramatic improvements in performance.

How is power wasted?

It is proven that the average Data Centre only uses 30% of it’s power load to operate the IT infrastructure, the rest is consumed by non-critical functions such as heating, cooling, lighting and resilient power systems. 99.9% of energy ultimately leaves the Data Centre as wasted heat.
 
on365 understand that only 5 key contributing factors need to be addressed:   

1. Inefficiencies of the power equipment
   Redundant architecture or devices operating below rated power make efficiency fall dramatically. What’s more, the wasted heat energy created by these devices only adds to the overheads and drop in efficiency in the cooling system.
   
2. Inefficiencies of the cooling equipment
   All cooling devices consumer power whilst operating and in doing so emit heat waste rather than add cooling to the Data Centre.
   
3. Power consumption of lighting
   All lighting adds heat to an environment and often consumes unnecessary power. Heat generated by lighting must be cooled by the cooling system, which causes a corresponding increase in power consumption by the air conditioning system.
   
4. Over-sized power and cooling systems
   Overestimated loads/risk analysis leads to excessive cooling/power capability (and cost)
   
5. Inefficient data centre configuration
   The overall configuration of a Data Centre can have a profound effect on the energy consumed by the cooling system. An ineffective configuration forces the cooling system to move excessive amounts of air and can also result in the cooling system generating more air than the Data Centre actually needs.
   

 An Optimized Data Centre Architecture
One of the easiest conclusions drawn from a review of these factors is that they are all interrelated. Using this knowledge on365 approach the planning of Data Centres with an overall focus rather than tackling each of the individual factors in isolation. This is by the far most efficient method. Careful analysis of the five contributing factors leads to the conclusion that Data Centre efficiencies can be substantially improved when an integrated system is developed, but only when based on the following principles:

• Power and cooling equipment should not be operated if it is not needed.
• Excessive power/cooling infrastructure is kept to a minimum.

on365 Design Principles
These guide the implementation of our systems:

• The latest cooling and lighting technology should be deployed wherever possible to minimise power consumption.
• Subsystems that must be used below their rated capacity (to support redundancy) should be optimised for that fractional-load efficiency, not for their full-load efficiency.
• Capacity management tools should be used to manage ”wasted capacity”  within the Data Centre, allowing the maximum amount of IT equipment to be installed within the constraints of power and cooling capability.
• An integrated physical configuration should be optimised and not tied to the characteristics of the room where it resides — for example, row-based cooling should be integrated with the IT racks, independent of room-based cooling.
• Load monitorig systems should be implemented to identify and warn of conditions that generate sub-optimal electrical load consumption, so that a quick fix is initiated and further loss avoided.

How does an on365 architecture differ from a conventional installation?
The possible reduction in electrical waste described above is dramatic. How does on365 achieve such results? What are the new technologies, designs and techniques that we use to increase efficiency so effectively? How can you measure our success? Some of the answers are in the specific elements of our design philosophy that bring together all of the improvements from a new architecture:

• Scalable power and cooling architecture to avert unnecessary capital investment and to avoid over-sizing.
• Row-based cooling to improve cooling efficiency.
• Highly-efficient UPS to improve power efficiency.
• Variable-speed drives on cooling devices to increase efficiency at partial load and in cooler ambient temperatures.
• Capacity management platforms to optimize power, cooling, and rack capacity.
• Optimise room layout for cooling efficiency.

To conclude, when addressed in isolation, some of these elements can deliver an efficiency gain however it is only when planned and implemented as an overall Data Centre strategy that the best results are achieved. We’ve highlighted the most likely symptoms and cures for high level efficiency gain however must point out that there are an infinite number of outside factors that could affect the results of Data Centre efficiency gains in your specific circumstances, some of these factors include:

• Dropped ceiling for air return in a traditional data centre.
• Unco-ordinated room-perimeter air conditioners fighting each other.
• Lack of hot-aisle/cold-aisle rack layout.
• Energy-efficient lighting.
• Powering air handlers from the UPS.
• Imbalanced sizing of power and cooling systems.
• Full dual-path air handlers.
• Full dual-path chillers.
• Packaged chillers or DX glycol systems.
• Shallow raised floor (0.5m or less).
• Large auxiliary loads (personnel space, network operations centre).
• Hot and/or humid climate.
• Very long coolant pipe runs.