Creating sustainable high-performance computing to drive the data industry forward

By on
Creating sustainable high-performance computing to drive the data industry forward
One of the most encouraging aspects of reducing the cost and environmental impact of data centres is the ability to democratise their adoption.
Photo by h heyerlein on Unsplash

Addressing the resources associated with data is essential to drive innovation.

In the past twelve months, the development and use of Artificial Intelligence (AI), machine learning and High-Performance Computing (HPC), has led to increased demand for high performance data centre facilities required to support these new technologies.

Data-intensive projects and advances that were once only possible to achieve through universities, research institutes and large corporations, are now accessible to a broad range of industries no matter the geographical location. However, while the technologies driving this accessibility continue to spark growth across sectors including space, agtech, and healthcare the data centre supply chain supporting this accessibility needs to ensure it evolves to sustainably support the forecasted growth.

Currently the global carbon footprint for data centres accounts for more than 2% of global carbon emissions (the equivalent of the world's entire airline industry), with the number expected to rise to 3.2% in 2025. By 2040, data storage is predicted to account for 14% of the world’s carbon emissions.

In order to facilitate the growth and reduce the financial and environmental costs of the data industry, more efficient technologies must be implemented to reduce the volume of natural resources required to service the sector. This is where advances in cooling technologies and other sustainable innovations are becoming increasingly important to these growth industries. 

The data industry is already embracing strategies that allow for the expansion of data centre facilities while reducing their overall consumption of resources.

Cooling systems are a fundamental element of any data centre or high-performance computer hardware. While currently, most HPC systems utilise cooling fans that use air to reduce the temperatures of their equipment. This technology has scarcely changed in several decades, despite cooling being a crucial element for the viability of the power-hungry data industry.

This is where new technologies such as liquid immersion cooling are driving a more sustainable future for data centres and high performance computing. The technology works by immersing servers in a dielectric fluid to deliver much more effective and energy efficient cooling, significantly reducing energy consumption and running costs.

As liquid immersion is inherently more energy efficient than air cooling, data centre facility providers adopting liquid immersion can deliver world-leading energy efficiency regardless of the location and climate. Data centres also benefit from improved hardware reliability, reduced facility maintenance and the opportunity for zero water consumption.

Without more effective cooling, HPC servers are unable to run at full capacity and the data facilities that house them require much more energy to keep temperatures cool enough for processors to deliver peak performance.   

The data shows no sign of slowing

With current forecasts estimating that up to 500MW of new data centre capacity will be brought online in Australia over the next 5 years, there is no sign of the industry slowing down. By converting even 20% of this new capacity to liquid immersion cooling, the Australian data centre industry would save around $250M AUD in capital expenditure as well as around 1.7 billion Kilowatt hours in energy consumption, an additional saving of around $50M a year.

As the sheer number and volume of data continues to rise, another factor that must be addressed is the emissions associated with HPC and the data industry. While there is no sign of a reduction in the volume of data, new technologies already exist that enable the data to be processed using 30 per cent to 50 per cent less data centre energy than running the same workloads with traditional equipment.  

As more sustainable technologies hit the market, it is possible to significantly reduce the electricity required to service a rack of HPCs while increasing the output per computing unit.

Data centres use significant natural resources including electricity for cooling fans, water for air conditioning and diesel for backup generators required to secure power supplies in case of blackouts. This is where new and more efficient cooling technologies are gaining traction as a means of rapidly reducing costs, water and energy consumption as the data industry looks to reduce its environmental impact and mitigate fluctuating energy costs.

Democratising AI and Machine Learning

One of the most encouraging aspects of reducing the cost and environmental impact of data centres is the ability to democratise their adoption. While the use of AI and HPC is growing rapidly across Australia, data centres and most of their suppliers are concentrated in large facilities in Sydney and Melbourne. As less resource intensive technology is introduced, smaller facilities will become more viable and regional centres will be able to build and access high performance computing, solving the tyranny of distance faced by smaller Australian cities and regional centres.

By improving the viability of diverse data facilities, the industry can democratise access to AI, machine learning and HPC to drive innovation and springboard Australia’s digital transformation.

Looking Forward: The future is bright for the Australian technology sector as the onset of new technologies continues to revolutionise the use cases and access to data across the country.

Jonathan Buchanan is Chief Customer and Commercial Officer at EngineRoom.

Copyright © BIT (Business IT). All rights reserved.

Most Read Articles


What would you like to see more of on BiT?
How To's
Photo Galleries
View poll archive

Log In

  |  Forgot your password?