How to stay green while running increased data centre workloads
Written by Nicole Cappella Thu 7 Oct 2021
As people and businesses spend more time online, data centre workloads are growing – with no sign of slowing down. Data centre owners and operators are faced with different, sometimes competing priorities – to scale data centre services to meet demand, while still maintaining progress toward environmental targets.
Plans to add capacity at an existing data centre may be limited by the amount of available space, preventing the easy solution: simply increasing rack density to put more servers into an existing footprint.
To stay green in the face of increased data centre workloads, owners and operators must take a multi-faceted approach. A comprehensive strategy should include:
AI can be used by data centres to optimize workflows, increasing infrastructure efficiency. Using AI, data centre operators can automate repetitive tasks and explore the potential of new data centre frameworks. For example, AI can be used to orchestrate workflows remotely across a network of small, interconnected edge data centres rather than centralizing workflows at a large, enterprise-level facility.
Ageing hardware can prevent a business from meeting increased workflow demands responsibly. Advancements in hardware design are being made so rapidly that even an aggressive refresh strategy can leave companies with obsolete equipment. A recent Forrester report found that “on average, 40% of server hardware deployed at company data centres is more than three years old. Companies are adding capacity to support emerging workloads, but they retain ageing hardware for four years on average, which is longer than ideal.”
To improve capacity through hardware while managing environmental goals, companies in the European markets have elected to:
- Upgrade server components 48%
- Select energy-efficient servers 39%
- Consolidate/eliminate lightly used or comatose servers 37%
- Employ greater server virtualization 37%
From 2019 to 2020, more companies globally plan to improve e-waste and recycling by partnering with an electronics recycling company, or recycling, reusing or repurposing decommissioned hardware.
Servers running more processes, at a higher rate, do so at a higher temperature – requiring more effort and resources to keep them cool. This is particularly true of data centres that attempt to increase rack density within an existing footprint, without adding to the facility’s physical space.
A recent study found that “the current average data centre cooling utilisation level is only around 40% – meaning that most organisations are using less than half of their cooling capacity.” Software-based optimisation approaches can improve utilisation while cutting consumption by up to 30% at the same time.
Innovations in data centre cooling systems can be useful both in helping data centres increase workloads and in limiting harm to the environment. Companies are exploring liquid immersion cooling as an alternative to air cooling, which can be more expensive to install and maintain, is less efficient in the use of both power and space, and carries a larger environmental impact.
Data centre owners and operators are under pressure from a number of different sources: from customers demanding increased reliability and almost no tolerance for downtime or service interruptions, from regulators with changing compliance requirements, and from the public at large who expect them to limit environmental impact. Balancing all of these priorities while still maintaining a profit for investors, providing innovative solutions, and growing the business can be a challenge. However, with a comprehensive strategy that covers workflow orchestration and automation, hardware upgrades and recycling, and innovative cooling and power systems, data centre operators have the opportunity to meet increasing capacity demands while still meeting aggressive environmental targets.
Written by Nicole Cappella Thu 7 Oct 2021
Tags:data center environmental goals green
AI Wed 20 May 2020Google to stop developing AI for oil and gas extraction...
Big Data Mon 1 Jul 201975% of all databases will live in the cloud by 2022
AI Thu 28 Nov 2019AI forces Go world champion to quit as it ‘cannot...