As Kermit the Frog famously sang, “it’s not easy being green,” yet it’s critical for our planet’s survival that we figure it out.

Ironically, the very tech we’re using to help in that effort could be making things worse. A 2019 survey conducted by Informa revealed that while datacentres use 3 percent of the world’s electricity, energy efficiency was only number four on IT leaders’ priority lists when creating a new datacentre. Yet those datacentres run the artificial intelligence and machine learning being used to understand and combat greenhouse gas emissions.

The price of AI
A 2019 report from the University of Massachusetts found that training and tuning a deep neural network can create the equivalent CO2 emissions of those produced in anything from just over two years of an average American’s life to more than seventeen years’ worth (depending on the model). Another study, also from 2019, found that the computations needed for deep learning research increased by 300,000x (no, that is not a typo!) between 2012 and 2018.

Going green
In an effort to combat the runaway emissions, at the Microsoft BUILD virtual conference last week, Microsoft, Accenture, GitHub, and ThoughtWorks announced that they have teamed up as founding members of the Green Software Foundation, a non-profit organization with the mission of creating an ecosystem of people, standards, tools, and best practices for building green software. It is part of the Joint Development Foundation Projects, an affiliate of the Linux Foundation.

What is green software
So what is green software? Sajay Podder, technology sustainability innovation lead at
Accenture explained. “The term green software relates to the design, development and use of software, and is a term we use to help companies consider the carbon emissions related to software with the intent to keep emissions as low as possible. It is important because while digital technologies are key to protecting the environment, that proliferation is also resulting in software’s growing carbon footprint,” he said. “But we also realize that it’s a new field, requiring new skills, new standards, new training and new tooling.” He said that when he discovered that a team at Microsoft was also looking at the issues, they began discussions about a collaboration, and the Green Software Foundation was born.

It’s all about the community, added Jeff Sandquist, corporate vice president of developer relations at Microsoft. Developers want to build apps that consider their impact on the

climate, and no one company or individual can do it alone. “For that to happen, we need to understand how to build green software, and folks need tools, and they need the knowledge to become qualified wherever they are in the world, on a really important matter,” he said. “And so our goal with this is to really establish best practices, set up standards, really work together. We’ll work together to be able to make sure that we create a great community of green software practitioners.”

Erica Brescia, chief operating officer at GitHub, said that GitHub joined with the goal to make it easier for developers to use and contribute to the global climate solution by empowering them with the right visibility and the right tools, so they can create and maintain green software themselves.

For ThoughtWorks, said Elise Zelechowski, global head of diversity, equity and inclusion, sustainability and social

change, sustainability and climate change mitigation is a priority. “We believe that the tech sector has a really important role to play here in taking action on climate,” she said. “We’re increasingly seeing our clients coming to us for thought leadership tools and advice on this. And while there’s a desire to take action, it’s not always clear where to start for them. ThoughtWorks is very excited to be joining the foundation to help define these standards to help organizations adapt and implement new tools, and to work with government, academia, nonprofits, and all of you and build this community to shape the vision of the work.”

What Is Green Cloud Computing?

Green computing—sometimes called green IT—focuses on reducing the environmental impact of computing infrastructure. It addresses energy consumption across the full stack: physical devices, cooling systems, network equipment, lighting, and the software running on top of it all. The goal is to use computing resources efficiently while minimising electronic waste and carbon emissions.

In the context of cloud computing, this translates into two broad areas:

• Software techniques: Developing methods and architectures that improve efficiency in storage, workload scheduling, and energy use at the application and platform level.
• Hardware optimisation: Designing and operating physical infrastructure—servers, network gear, cooling—to minimise energy consumption and support recycling and responsible disposal of components.

Why QA Engineers Need to Care

Modern QA environments host a wide variety of workloads: short-lived unit test runs, long-running integration suites, parallel browser-based tests, and always-on monitoring agents. Managing these diverse workloads on shared cloud infrastructure creates real challenges around provisioning, allocation, and idle resource waste.

QA practitioners who understand green computing can:

• Design test pipelines that scale down when not in use, reducing idle compute costs.
• Choose cloud regions or providers with lower-carbon energy mixes.
• Consolidate test infrastructure through virtualisation rather than running dedicated physical machines.
• Reduce overall CO₂ emissions as a side effect of writing leaner, faster tests.

Key Green Cloud Computing Techniques for QA Teams

Energy efficiency in the cloud can be approached from both the infrastructure layer (managed by cloud providers) and the workload layer (influenced directly by QA teams). The following techniques are the most relevant for QA practitioners to understand and, where possible, influence.

1. Virtualisation

Virtualisation is the foundational green technology in cloud computing. A hypervisor allows multiple virtual machines (VMs)—each with its own operating system and application stack—to share a single physical server. This dramatically improves hardware utilisation rates compared to running one workload per physical machine.

For QA teams, virtualisation means:

• Test environments can be spun up and torn down on demand, avoiding always-on infrastructure.
• Multiple isolated test environments (different OS versions, browser configurations, dependency sets) can run on shared hardware.
• Dynamic migration allows VMs to be moved to more efficient hosts as workloads shift, improving energy use across the cluster.
• Idle VMs can be suspended or terminated automatically via auto-scaling policies, cutting wasted compute cycles.

Container-based approaches (Docker, Kubernetes) extend these benefits further by packing workloads more densely and starting faster than full VMs, though the underlying principle—maximising resource utilisation—is the same.

2. Dynamic Voltage and Frequency Scaling (DVFS)

Dynamic Voltage and Frequency Scaling (DVFS) is a processor-level technique that adjusts CPU clock speed and supply voltage in real time based on current workload demand. When a processor does not need to run at full speed, DVFS reduces both frequency and voltage, cutting power consumption significantly—because power scales roughly with the square of voltage.

In practice, DVFS is managed by the operating system and hardware firmware rather than directly by QA engineers. However, QA teams can influence it indirectly by:

• Choosing cloud instance types that are right-sized for the actual workload, allowing the CPU to spend more time at lower frequency states.
• Avoiding test designs that create artificial constant-load conditions (such as tight polling loops) that keep CPUs pinned at maximum frequency unnecessarily.
• Using burst-capable instance types for intermittent test workloads rather than high-performance instances that run at full power continuously.

3. Nano Data Centres

Nano data centres are a distributed computing model that challenges the conventional approach of building large, centralised data centres. Instead, the concept distributes computing capacity across many smaller nodes located geographically closer to end users—similar in principle to edge computing architectures.

The potential energy advantages include:

• Reduced network transmission distances, which can lower overall energy consumption in the delivery path.
• Smaller facilities that may be easier to power with local renewable energy sources.
• Lower cooling overhead compared to hyperscale data centres in some configurations.

For QA teams, this concept is most relevant when testing geographically distributed applications or latency-sensitive services, where choosing edge or regional test infrastructure reduces both test latency and the energy cost of long-haul data transfer.

4. Auto-Scaling and Workload Scheduling

One of the most practical green cloud techniques available directly to QA teams is intelligent workload scheduling combined with auto-scaling. Rather than keeping test infrastructure running around the clock, teams can:

• Schedule long-running test suites to run during off-peak hours when data centre load is lower and energy may come from a cleaner grid mix.
• Use auto-scaling groups or ephemeral CI/CD agents that provision nodes at test start and terminate them on completion.
• Consolidate parallel test jobs onto fewer, fuller instances rather than spreading them across many lightly loaded nodes.
• Power down non-production test environments overnight and on weekends.

Hardware and Software Optimisations in Practice

Green cloud computing can be broken down into hardware-layer and software-layer optimisations. QA engineers typically operate at the software layer but should understand both when evaluating infrastructure choices.

Hardware Layer

• Servers, network switches, and storage arrays in the data centre are the primary energy consumers.
• Cooling systems (HVAC, liquid cooling) can account for a significant proportion of total data centre energy use—Power Usage Effectiveness (PUE) is the standard metric used to measure this ratio.
• Hardware-level techniques include powering off idle components, using energy-efficient processors, and optimising physical layout for airflow.

Software Layer

• Cloud management systems orchestrate workloads across physical resources to maximise utilisation and reduce idle servers.
• At the application layer, techniques include reducing CPU polling, using event-driven architectures, optimising database queries, and minimising unnecessary network calls—all of which reduce energy consumption as a side effect of better code.
• In the compiler and operating system layers, techniques like slowing CPU clock speed under light load, improving performance per watt, and shutting down idle cores contribute to overall efficiency.

The Four Pillars of Green Computing

Green computing is generally organised around four areas of action. QA teams can contribute to each:

1. Hardware device manufacturing: Prefer cloud providers and hardware vendors with published sustainability commitments and responsible supply chains.
2. Software techniques: Design test pipelines and test code to be lean, fast, and resource-efficient.
3. People awareness: Educate QA teams about the energy cost of always-on infrastructure and encourage habits like terminating unused environments.
4. Standard policies: Advocate for organisational policies that enforce auto-shutdown of non-production environments, right-sizing of cloud resources, and carbon-aware scheduling.

Getting Started: Practical Steps for QA Teams

• Audit your test infrastructure: Identify always-on environments that could be made ephemeral.
• Enable auto-scaling on CI/CD agents: Use cloud-native or self-hosted runner pools that scale to zero when idle.
• Right-size your instances: Match instance types to actual CPU and memory requirements rather than over-provisioning.
• Measure before optimising: Use cloud provider cost and usage dashboards to understand where compute is being spent, and use that data to prioritise changes.
• Choose greener regions: Many cloud providers publish carbon intensity data by region; running workloads in lower-carbon regions is a low-effort, meaningful change.

Summary

Green cloud computing is not just an infrastructure concern—it is increasingly a shared responsibility across engineering disciplines, including QA. By understanding techniques like virtualisation, DVFS, nano data centres, and intelligent workload scheduling, QA engineers can design more sustainable test pipelines, reduce wasted cloud spend, and contribute meaningfully to their organisation’s environmental goals. Efficiency and sustainability, in this case, go hand in hand.