On the Way to Green Cloud Computing: How to Balance Scalability with Sustainability

Putting a figure on the cloud’s environmental impact

According to this MIT research, the cloud’s carbon footprint has already exceeded the airline industry’s. 

 

An individual data centre can use as much electricity as 50,000 households. Data centres consume 200 Terawatt hours (TWh) of energy yearly, which is more than some nation-states. 

 

Apart from electricity, the cloud is also thirsty. Another study estimates that a large data centre can ‘drink’ between 1 million and 5 million gallons of water a day (3,8 – 18,9 mln litres). That’s as much as a town of 10,000 to 50,000 people!

 

Of course, you can’t impact your hyperscaler’s global moves as an individual user. However, your cloud infrastructure decisions can make a huge difference regarding your footprint. 

 

Here’s what to consider when designing a green cloud system for your business. 

Five industry best practices for green cloud computing 

1. Choose regions based on business and sustainability goals.

This best practice refers to selecting cloud resources necessary for your business KPIs from regions marked as renewable energy projects or locations with low carbon intensity. 

 

You can use resources like Electricity Maps to determine which areas have lower carbon footprint levels. 

 

Cloud providers like GCP also offer tools that help you pick the most optimal regions for your requirements based on carbon footprint, price and latency. 

 

2. Match your demand dynamically. 

This practice involves scaling resources against load. You can avoid adding redundant resources by testing and configuring cloud resources against demand and ensuring flexibility in this matter. 

 

Scaling – dynamically adding and removing resources for your workload – positively impacts costs and energy consumption in server rooms. 

 

Moreover, you can use software optimisation and event-driven architecture to plan the use of resources according to the situation. 

 

Serverless solutions like AWS Lambda are great for green cloud computing systems as they only use as many resources as allocated for a given activity.

 

3. Optimise software and architecture.

It’s possible to design a cloud system to make optimal use of reserved resources. 

 

Implementing appropriate technologies for smooth operation and keeping the load on components at a high level of resource consumption reduces their idleness from lack of use. 

 

It can be implemented with multi-threading implementations, for example thanks to the use of goroutines in the Go language or threading in Python. 

 

Allowing multiple threads to execute simultaneously on a single processor helps to ensure its more efficient use. This feature, in turn, reduces the processor’s idle time waiting for a single thread to finish, and less idle time is lower power consumption. 

 

4. Improve your data storage and management practices.

Suppose you reserve excess storage for not-yet-existing data or keep it longer than necessary. In that case, your vendor may want to increase the pool of resources available to customers and, therefore, consume more natural resources.

 

Modern technologies offer many data storage services for varying amounts of data, needs, and purposes. Data lifecycle is an essential aspect for your infrastructure architect to consider to limit the amount of redundant data space in use. 

 

Doing so ensures appropriate data classification by determining if you need fast access for long data storage or if you can retrieve it slower. And ultimately, it allows you to choose the right energy-efficient tier for your data. 

 

5. Make the right resource and service decisions.

This green cloud computing standard relates to keeping the amount of hardware necessary for implementing the system or application to a minimum. 

 

Choosing the most appropriate hardware, like an instance type or service for a workload, impacts sustainability positively as it helps to reduce waste. Moreover, carefully selected energy-efficient instances can significantly reduce your expenses and emissions. 

Marching towards a more sustainable future

Lower hardware use and lower resource utilisation translate to lower energy consumption. 

 

With over 300 new sites in the pipeline, the number of hyperscale data centres will likely exceed 1,000 by the end of 2024. All these corporations and their data centres already require massive amounts of electricity to match their customer demand – and this level will most likely only grow. 

 

Cloud providers like Amazon and Microsoft have declared switching to use 100% renewable energy in their data centres by 2025, while Google made such a promise by 2030. Additionally, all major vendors focus on carbon, water, waste, and the ecosystem, as their customers and public opinion demand them to do so. 

 

As CSPs commit to and achieve their renewable energy goals, the momentum towards greener computing will grow. By choosing relevant tools and practices, your business can join this collective effort and, hopefully, contribute to curtailing the need to expand data centre facilities.