Matthew Margetts – Director of Sales and Marketing at Smarter Technologies
Inside vast factories bigger than aircraft carriers, tens of thousands of circuit boards are racked row upon row. They stretch down windowless halls so long that staff ride through the corridors on scooters. In an increasingly digitalised world, data centres are the information backbone with demand continuing to grow along with data-intensive technologies. Estimated to account for as much as 1% of worldwide electricity use, data centres are energy-intensive enterprises.
In Ireland, data centres could account for about 25% of the country’s electricity usage by 2030, potentially leading to electricity supply challenges. Fearing the pressure that data centres place on the national grid, countries such as the Netherlands and Singapore have gone so far as to stop issuing building permits to data centres.
Why do data centres require so much energy?
– To provide constant power supply with minimum disruptions
– Electricity used by IT devices such as servers, storage drives and network devices is converted into heat, which must be removed from the data centre by cooling equipment that also runs on electricity
– Facilities must be kept at the appropriate temperature
– Additional equipment such as humidifiers and monitors are also required
The energy impact of data centres is undeniable but so is the need for these facilities to handle the world’s ever-increasing data demands. What can’t be ignored is the energy efficiency trends that have developed in parallel. The IEA reports that although workloads and internet traffic have nearly tripled, data centre energy consumption has flatlined for the past three years.
Here’s what can be done to improve data centre energy efficiency and sustainability:
The use of server virtualisation and ARM-based processors can help reduce the energy consumption of IT devices. This new technology is designed to perform fewer types of computer instructions, allowing them to operate at a higher speed and resulting in better performance at a fraction of the power. The servers of today are more powerful and efficient than ever before and the technology continues to improve.
One of the best ways to match the rise in ICT workload energy is to ensure a corresponding increase in the usage of renewable energy sources.
By moving part of their high-intensity computing hardware to alternative locations using renewable energy, companies can benefit from a more sustainable energy source while taking energy off the national grid. A location like Iceland boasts reliable, low-cost renewable energy.
Big data centre operators such as Google are establishing solar generation plants to offset their data centre usage on the grid, using small panels coupled with battery storage to reduce non-critical functions such as engine heaters, office air-conditioning, fuel polishing and lighting.
Intelligent power distribution management
The key to better energy efficiency in data centres is managing power load and distribution. For example, reducing the number of servers needed during low traffic hours. Rather than leaving all servers idle, some servers can be turned off when not needed while others run at full throttle. Matching the server capacity to real-time demands is made possible through smart monitoring and management tools.
It’s also important to remove “zombie servers” which are servers that have become redundant and are no longer in use, yet are still powered on and consuming energy. Research shows that 25% of physical servers are zombies, along with 30% of virtual servers. In general, these servers haven’t been shut down because operators don’t know what they contain or what they are used for. To deal with this problem proactively, every server and function must be documented and monitored appropriately using asset management software.
In conventional data centres, standard air conditioning uses a significant proportion of the centre’s energy bill. All IT equipment must remain at safe temperatures which is why proper ventilation and cooling is so important.
Measures managers can take to optimise cooling include the following:
– Proper insulation can help maintain temperatures within the room.
– Strategic equipment layout and streamlined airflow can also improve cooling efficiencies.
– A popular solution is to locate data centres in cool climates and use the outside air to cool the inside. This is known as “free cooling”.
– Piped water is a good conductor of heat. Warm water can be used as a less energy-intensive way to cool data centres.
– Cleaning up workloads and eliminating unnecessary equipment.
– Replace older cooling systems with new technology to improve efficiencies.
Machine learning and automation in data centres can also be used to optimise cooling system setpoints for variable outside conditions which provides a number of marginal energy gains.
Heat transfer technology
Using the heat coming off the servers is like taking advantage of a free resource. For example, an IBM data centre in Switzerland warms a nearby swimming pool with its waste heat.
However, because heat doesn’t travel well, the use of waste heat is generally limited to data centres that can supply nearby customers or cities that already use piped hot water to heat homes.
The information age is making buildings smarter and more energy efficient. With fairly simple automations such as occupancy sensors that turn off lights and HVAC when no one is in a room, along with informed decision-making as a result of access to real-time utility consumption data, building managers can use smart technology and building management systems to reduce their carbon footprints. This infrastructure is facilitated by data centres, so one could argue that some of the energy being used by data centres is offset by the lower consumption of the smart buildings they service.
Policy making and planning
Decision-makers need to be able to confidently and accurately evaluate future efficiency and mitigation options.
Policymakers and energy planners need to be able to:
– Monitor future data centre energy use trends
– Understand key energy use drivers
– Assess the effectiveness of various policy interventions
In order to do this, data analysts need access to reliable data sources on the energy consumption characteristics of IT devices and cooling/power systems.
Smart metering technology is just the start—along with the data from smart meters, energy managers need a platform with data analytics, artificial intelligence and machine learning capabilities in order to make the most of the data they are presented with.
Data centre operations require a safe, efficient and dependable power supply. There’s no doubt that sustainability is going to be the overriding trend that will remain front and centre within the data centre industry for the foreseeable future. Fortunately, the very same smart technology that is necessitating the growth of data centres is also helping to make them more energy efficient and future-fit.