Low-carbon Hosting

Description

Pre-compilation, the technique of pre-compiling, optimizing, compressing and caching resources before they are served to the user has become a key benefit of Jamstack and similar architectural approaches. Platforms such as Netlify offer a variety of pre-compilation options, not just for static-site generators and markups, but also images and other media. The result is the transparent optimization and distribution of resources for the developer without any special built configuration.

What's the impact?

At its simplest, precompilation helps to reduce the payload size and request process time. However, platforms are starting to differentiate on their precompilation options, offering more intelligent compression, caching and CDN automation to further lower your site or app's carbon footprint.

Description

Cloud computing, in certain applications and configurations, can provide meaningful carbon benefits to organizations. When evaluating cloud providers, the sustainability activities of that provider should be taken into consideration as part of the procurement process. The leading cloud providers—Microsoft Azure, AWS, Google Cloud Platform and IBM Cloud—all provide satisfactory levels of commitment to sustainability, with benefits and services for organizations in pursuit of their own sustainability agenda. While some providers are further ahead than others, all have some level of commitment which can be evaluated against your own organization's objectives.

What's the impact?

This pattern is something that frog routinely employes for the products and services that it builds. We work closely with the expertise across Capgemini to ensure we're regularly aligning with best practice in the cloud space and choosing the right platform for our needs.

Learn more

Read more about each cloud platform's sustainability objectives:

• Microsoft Azure
• Google Cloud Platform
• AWS
• IBM Cloud

Description

A Content Delivery Network (CDN) is a geographically distributed network of proxy servers and data centers that enables efficient distribution of data and assets. It speeds up data retrieval for end-users by fetching data that is geographically closest to that user and distributing load across the network of edge-serves to mitigate service interruption.

What's the impact?

CDNs decreases network traffic by distributing the service spatially relative to end users. This reduces the energy required to serve applications and content, and often prompts developers to consider what can be cached and distributed as part of the development process. frog routinely employes CDNs as part of the products and services that we build. It's commonly adopted as an industry standard technique, and we strongly advocate its use.

Description

Serverless deployment is now a mainstream offer from all cloud providers. It covers various means of deploying and serving apps/services from the cloud in an on-demand way.

What's the impact?

Serverless deployments are a resource-efficient way to deploy apps and services without procuring dedicated cloud resources. Your app or services are only served when needed and have minimal footprint when not in use.

Description

Consider running resource-intensive services or workloads in regions with a lower carbon intensity—such as opting for more remote locations rather than built-up areas. There are various ways to identify this; ElectricityMap is a website which details the energy mix for various locations.

What's the impact?

By shifting hosting to an area with less demand on resources, it's possible to reduce the environmental impact of resource-intensive processes. Locations with higher percentages of renewable (and low-carbon) energy can be identified and therefore we are able to estimate what percentage of energy use is sustainable. Carbon impact is just one of the many wider considerations that must be evaluated when choosing cloud service regions. While it may not always be possible, the carbon impact of a region should be a metric that is evaluated as part of a deployment plan.

Learn more

• electricitymaps.com

Description

Platform as a Service (PaaS), such as Netlify, Vercel and Google Firebase, provide turn-key services for the development and distribution of mobile and web applications. They often bundle in a number of quality-of-life features and simplifying common development concerns such as compilation, hosting and CDN integration as well as authentication, data management and compute functions.

What's the impact?

These platforms are often built atop one or more cloud providers and inherit their carbon footprint and sustainability agendas. Google Firebase, for instance, leverages Google Cloud infrastructure and benefits from Google's strong suitability focus. Similarly, these platforms often lead developers to follow methodologies or use features that are recommended in this Radar—such as demand shaping, precompilation and asset optimization and abstracting microservices. For certain applications, notably mobile and web applications, these platforms offer a compelling sustainability advantage. However, it is always important to evaluate each platform's individual sustainability agenda.

Learn more

• Netlify & Sustainability
• Vercel's Green Energy Policy
• Google Cloud Sustainability

Description

Evaluate providers with stated policies and goals that work toward more sustainable and responsible technology.

What's the impact?

Smaller, sustainability-focused hosting operations can be considered risky. However, when selecting a supplier of cloud services, we should always consider the sustainability credentials of the supplier and, in certain use-cases, the sustainable difference of these providers can offer a compelling differentiator.

Learn more

• The Green Web Foundation's directory of verified green hosting providers

Description

Maintaining your workload distribution and computing resources means that you can use fewer resources at a higher utilization. This can be done by either decreasing the size of your resources to reduce waste or redistributing compute to take advantage of any unused capacity.

What's the impact?

When deploying a compute resource in the cloud, it is most energy efficient to use as much of the available compute as you can before deploying additional resources. Meanwhile, where you have unused compute, you can consider reducing the size of the instances to be more in line with utilization. Using auto-scaling and burst functionalities of cloud platforms can allow you to increase and decrease your compute-resources based on demand while keeping the utilization high. It may be that your hosting solution doesn't provide for this level of access, however, where possible, this technique should be explored.

Description

The data-center model, whether an on-premises or tenanted model, is difficult to align with an organization's sustainability agenda. These patterns tend to require dedicated hardware and lead to over-provisioning. Whenever data soveriegnity or regulatory requirements do not constrain an implementation, migrating to the cloud will be the most sustainable option.

What's the impact?

Data-centers represent a large portion of the overall energy cost of the internet, and with the prevalence of cloud computing and PaaS, there are fewer needs for organizations to utilize their own on-premises or tenanted data-centers when sustainable cloud options are available.

Description

Relying on pre-allocated, dedicated or fixed capacity or usage on a cloud platform, means not taking full advantage of cloud capabilities, such as scaling on demand. This would lead to over-provisioning, which will lead wasted resources and spending.

What's the impact?

Over-allocating fixed resources in a cloud platform uses energy, even when there is no demand for those resources. Ideally, organizations should configure only for what is needed, and utilize the scaling features of your cloud infrastructure to scale with demand. In many applications, pre-allocation should be avoided for both its cost and environmental impact. However, in situations where usage remains relatively stable so as to minimize under-utilization—pre-allocation may present as a suitably sustainable route.

Description

Cloud platforms and services that don't frequently upgrade their tech stack support are missing out—using the latest stable version of web technologies will generally deliver the best performance, security and energy efficiency.

What's the impact?

The latest server tech stacks are becoming increasingly efficient, and are able to achieve the same outcomes with less processing and therefore less energy. We should prioritize platforms that are upgraded frequently to run the latest versions of services and stacks.