Re-thinking data centres


27 June 2014
David Blumanis of Schneider Electric

The way data centres are designed, built and managed will have to change in order to address issues such as time to market and energy consumption as they compete in the battle for Internet of Things (IoT) and cloud computing.

Speaking at a media meeting in Singapore, David Blumanis, vice president of Data Centre Solutions and Key Account Management, Asia Pacific and Japan, Schneider Electric IT Business, noted that data centres are becoming more mission critical as data volumes explode, consumer expectations rise with respect to performance and reliability, and IoT gains traction. 

“IoT is just around the corner,” he said. “We are only scratching surface right now in terms of the amount of data that will be created by the Internet of People and in the future, the Internet of Everything.”

Closely tied to this is the concept of smart cities, where fast-growing populations will demand data-driven solutions, underpinned by IoT, to address urban efficiency challenges such as utilities management, water management and traffic congestion.

The energy conundrum

Data centres play a critical role as the brain of a smart city, and extremely large facilities are being planned in this region to cater to the data that will arrive. However, some of these “data motherships” could have problems taking off due to a power crunch, observed Blumanis. “These data centres cannot be built in some cities because of their power utilisation requirements.”

Data centres are important consumers of energy and will become even more important in the future, said Philippe Arsonneau, senior vice president, Global Sales, Schneider Electric IT Business. “An average of 5-6 per cent of energy consumed in a country today is consumed by data centres and this will grow to 15-20 per cent by some crazy calculations. There is a big growth in energy requirements because of IoT, because of the growth in data exchange, because machines are being connected to machines,” he said.

The energy conundrum is not being helped by a rather laissez faire attitude towards data centre power consumption. Simon Piff, associate vice president, Asia Pacific Enterprise Infrastructure Research Group, IDC, presented findings from the IDC Asia Pacific Transformative Infrastructure Index 2014 where nearly 40 per cent of respondents in the region said they were not accountable for power and cooling costs.

This was in response to the question as to whether they measured and cared about power consumption in their data centre.  Over one third of respondents said they had no specific view of energy consumption in their data centres, while less than a quarter said they were mandated to control the power and cooling cost.

This presents a big missed opportunity for cost savings in data centres. “One of largest opex costs on the books today is electricity,” said Blumanis. “It has been escalating and will continue to escalate.”

It is estimated that a green data centre, optimised for energy efficiency, can yield savings of up to 30 per cent.

Integrating energy efficiency into data centre management practices

One approach to achieving a green data centre is to integrate energy and environmental efficiency into data centre management practices. According to Keith Murray, vice president, Singapore and Brunei, Schneider Electric IT Business, this involves critical environment assessment, base-lining and target setting; green data centre certification consultancy; metering and monitoring; and continuous improvement.

A detailed assessment of the data centre is carried out based on metrics defined in a green data centre standards such as Singapore’s SS564, which defines policies, systems and processes for “going green”. This establishes the baseline condition of the data centre for future comparison and allows targets to be set for future energy performance.

Green data centre certification consultancy involves the preparation of the environmental management system - scoping out the system, describing its main elements, setting out the green data centre policy, objectives and targets and drawing up the action plans to achieve these targets.

Metering and monitoring systems have to be implemented so that data centre operators can make informed decisions with complete visibility into the data centre’s physical infrastructure. These systems would include a custom dashboard, tailored reporting based on the green data centre standard and individualised data tracking to provide insights into the key performance indicators for the data centres.

Schneider Electric also advocates a continuous improvement programme where data centres identify opportunities to keep improving energy efficiency and take a more proactive approach to building a sustainable, energy-efficient foundation for their operations.

Adopting a more holistic approach to data centre design

On the infrastructure front, Schneider believes that there is a limit to how far energy optimisation can be achieved at the component level. “We can’t get more efficient, or more reliable, just by building better components,” said Blumanis.

Instead, it is suggesting an evolution in approach - from the use of individual components to the deployment of pre-assembled, factory-tested subsystems for the IT room, facility power, facility cooling and security, delivered with data centre infrastructure management and data centre lifecycle services.

Compared with traditional design and build method which involves bringing in different vendors for different infrastructure components and systems, going through a long design process and complicated build schedules, the modular approach will result in decreased installation time and design effort, and allow data centre operators to grow infrastructure in whitespace on demand. At the same time, pre-testing in the factory also translates to improved reliability.

Taking this proposal one step further, Schneider is also looking to deliver modularity in larger building blocks with pre-fabricated, modular data centres that can deliver the agility needed to keep up with data growth, said Blumanis.

With such an approach, the entire system can be designed to work together to achieve maximum PUE (power usage effectiveness). It also simplifies data centre infrastructure management by paving the way for centralised management of integrated sub systems. There will be lower installed costs, lower TCO and better performance, and deployment can take as little as 10-12 weeks, he said.