24/06/11
By Jason Preston, Director of data centre innovator, 2bm
The continuing need for cost containment combined with new government legislation is placing greater pressure on Data Centre managers to drive down power consumption and improve efficiency. Indeed, new state of the art data centres are achieving phenomenal efficiency ratings through the use of innovative technologies such as direct and indirect free cooling, with a strong emphasis on recycling and sustainability. But just how applicable is this best practice to existing data centres?
Applying Best Practice
With the introduction of the Carbon Reduction Commitment (CRC) Energy Efficiency Scheme targets, the government is demanding organisations make tangible commitments to reducing energy consumption. But with the drive to meet sustainability targets and drive down costs, CIOs are being tasked with significantly improving this rating and moving closer to efficiency nirvana - namely a Power Usage Effectiveness (PUE) of 1.0.
The evolution in data centre design and innovation is continuing at a phenomenal pace. Capgemini’s new state of the art Merlin facility, for example, takes data centre thinking to a whole new level; indeed its innovative design and exploitation of leading edge technology has been rated as functioning at an amazing 1.08 PUE.
Understanding Data Centre Performance
Of course, few organisations have the luxury of tearing down the existing data centre and building a new one from scratch. The good news is that there are a number of steps that can be taken to improve the performance within the existing set up and maximise efficiency — from considering how air comes into the data centre to how it is cooled, how racks are arranged and where cables are run.
But before any investment is made, organisations must understand the current levels of performance. So what are the options for improving the PUE of existing data centres?
Ten Top Tips for an efficient data centre
1) The most obvious step is to configure cabinets in hot and cold aisles. This is a fundamental design, without which it is extremely difficult to leverage any of the other technology innovations or implement design best practice that can make a significant difference to data centre performance.
2) Avoid the mixing of cold air supplied from the air conditioning systems with hot return air from the IT equipment. For example, ventilated floor tiles need to be correctly positioned within the cold aisles and preferably be fitted with VCDs (volume control dampers)
3) Fit blanking panels to any cabinet free U space and fit brush sealed grommets or blanks to any holes within a raised access floor, if it is being used as a supply air plenum.
4) Install hot or cold aisle containment systems. Controlling every aspect of the supply air path from an air handing unit, through the servers and back, prevents recirculation and mixing of hot and cold air. This improves the operational efficiency of air conditioning systems. Both hot and cold aisle containment systems can achieve this benefit.
5) Increase the room (or cold aisle) temperature to between 24˚C and 27˚C. Understanding of this is now becoming more widespread, however too many data centre managers still think that their room should be set at a temperature of below 19˚C. The ASHRAE guideline is 18˚C - 27˚C for server inlet temperature and it is now generally accepted by manufacturers that servers can run far hotter without affecting performance or resilience.
6) Install indirect or direct free cooling. The practicality of these technologies is very site, and budget specific, using specialised free cooling plant externally installed coupled to traditional chilled water or DX gas CRAC systems. This utilises the outside ambient air temperature to transfer cooling to a water and glycol mixture piped to the indoor units via a closed circuit. Improvements in the efficiency of data centre cooling of between 40-70% is not uncommon.
7) Install high efficiency UPS systems and consider using UPS’s in smart or line interactive modes. New models of UPS are generally 96% efficient in online mode with output power factors of 0.9, and in some case 1. Many will have line interactive or ‘Smart’ modes where the inverter is turned off until the UPS senses an issue, or disruption, to the power supply.
8) Install voltage regulators and/or power factor correction equipment. The mains power supplied to many sites in the UK is either over the voltage required for the equipment installed or the power factor is low. The use of voltage regulators will stabilise the voltage being supplied to the data centre at a defined level. Power factor correction equipment will move the power factor as close to 1 as possible within reasonable cost. Both these technologies can deliver power cost savings of between 10-20%
9) Look at the lighting within the data centre. Wherever possible change lighting to high efficiency types with either compact fluorescent T5 or LED lamps and electronic control gear. Use proximity switching instead of standard switches so that once the area is vacated the lights will automatically turn off.
10) Replace legacy servers and storage with new high efficiency equipment; for example, leverage virtualisation to maximise performance of the equipment you already have.
Conclusion
Achieving an effective data centre design is a continuous and holistic process, with CIOs tasked with balancing cost cutting and CRC targets against business demands for more data centre capacity. Making the right decisions will therefore depend on detailed insight into existing performance, combined with the ability to model the impact of proposed changes to assess investment value. It is therefore essential to understand data centre performance before embarking on any changes. Using real time insight into cabinet temperatures, humidity and energy consumption will provide both a strategic overview and effective day to day management going forward.
