Through out the UK server rooms and datacentres face unprecedented pressures related to energy usage. The first issue is one of securing a reliable power supply and the second is one of how to use power most efficiently within the computing environment.
The UK national grid has traditionally had a high capacity of reserve supply. In the past, this safety margin was as a high as 15% of the overall power generation capacity. Sudden cold spells leading to increased demand for power for heating and lighting could be easily met. This is not the case today due to a reduction in overall generating capacity and rising demand across the country. Less carbon-friendly power stations have or are in the process of being decommissioned in favour of more environmentally friendly solutions including renewable power (solar PV and wind turbine) and nuclear power. At the same time, electricity demand is rising drive by population growth and the UK’s growing position as the centre of the European datacentre landscape.
Inside a server room or datacentre one of the key operational costs is electricity and its use in systems used to manage and control the environment; namely cooling and uninterruptible power supplies. To achieve the lowest possible electricity usage requires server room and datacentre designers to look at each critical system within their facilities and how to optimise their energy efficiency.
There is much talk in the press of running ‘green’ server rooms and datacentres. The first step on this path is to monitor energy usage, preferably against benchmarks and then to investigate ways to improve underlying system efficiencies. There are two key metrics that could be used for this.
DCiE looks at Data Centre infrastructure efficiency in terms of the percentage of input power for the IT loads and separating this from the power required for infrastructure systems including cooling, access, security and lighting. The formula for monitoring DCiE relates to a second measure used for monitoring energy usage known as PUE which stands for Power Usage Effectiveness.
Power Usage Effectiveness (PUE) = 1/DCiE
Power Usage Effectiveness (PUE) = IT Power/Total Power
Either formula can be used, with most datacentres now opting for PUE or some other derivative of this. The PUE ratio is also one used by colocation datacentres in the marketing of their services. The closer a PUE is to 1 (unity) the more efficient it is leading to potential lower operating costs for the datacentre and electricity bills for colocation users.
Energy efficiency can easily drop within a server room or datacentre environment and it is important to constantly monitor energy usage. Smaller Edge computing server rooms and micro-datacentres face the same issues when it comes to energy efficiency as mega-sized datacentre operations. The issue is one of maximising floor densities to ensure that the datahall / server room can house the required IT capacities within the right environmental in terms of ambient and power security. Consideration therefore needs to be given to:
Future expansion is important as IT technology continues to deliver ever more powerful servers. The ability to easily scale up or down must be considered for each critical system in order to maintain optimal efficiency.
In terms of UPS systems scalability (both horizontal and vertical) is possible using the latest high efficiency modular UPS systems. Traditional transformer-based UPS systems can incur high standing losses in terms of the power used to energise their transformer and associated electronics. Wasted energy is evident in terms of heat (transformer hum and cooling fan noise) and heat emissions. The heat adding a further burden to the cooling required within the room. Transformerless UPS systems are more energy efficient and when combined with modular UPS technology a scalable architecture can be deployed.
Modular UPS systems use a frame structure with expansion bays. A typical 300kW frame may use up to six 50kW UPS modules. The frame can be installed with three 50kW UPS modules to give 100kW N+1 redundancy and still allow up to a further three 50kW UPS modules to be added at a later date. With a modular UPS system, the on-line operating efficiency can be around 96% with a flat load curve allowing this energy efficiency to be maintained over a wide load range and often down to as low as 25%.
It can be argued that cooling costs are the biggest contributor to energy costs within a server room or datacentre environment. There are several competing designs to choose from dependent upon the size of the datacentre, location and access to free cooling and of course the configuration within the data hall(s) themselves.
Whichever cooling method is chosen the focus is on maintaining the right ambient environment in terms of cooling and humidity and the overall energy efficiency of the cooling design. This can be measured using the following formula:
Cooling System Efficiency = Useful Cooling Power / (Power Consumed + Useful Cooling Power)
Cooling efficiency is a useful measure when deciding how to cool the IT environment with hot aisle/cold aisle arrangements often being used to achieve optimal energy efficiency. Rear door cooling can also be added to server racks to improve energy efficiency.
Many organisations are now employing energy management systems (EMS) in their drive to become more energy efficient. It may be possible to monitor energy usage for infrastructure systems like UPS and cooling for server rooms using an EMS. For large and mega-sized datacentres, the use of Data Centre Infrastructure Management (DCIM) software, may be more appropriate. Whether you are looking at a new build server room or datacentre refresh, the consultants at Server Room Environments can help optimise your power usage effectiveness and cooling system operations.
At Server Room Environments we see a growing importance for on-site server rooms, even with more businesses turning to Cloud based services. What’s driving the demand for on-site IT facilities is the need to prepare for the Internet of Things, Edge Computing and the level of critical infrastructure that organisations will need soon.