Cloud principles

Principle 6: Be efficient

Statement

Use cloud services in a cost-efficient and environmentally sustainable manner.

Why?

Using or developing services that run lean and scale on-demand reduces the cost and environmental impact of running the service.

Services designed with efficiency in mind usually are less complex, which reduces the burden of ownership and maintenance.

What does this mean?

• design services to run lean and scale to meet demand automatically. Service resource utilisation should be as high as possible, without impacting desired performance levels
• focus on designing your service to fit your chosen cloud delivery model (Principle 1). Avoid applying existing architectural approaches to cloud delivery, as it increases costs and complexity while removing many of the benefits of public cloud services
• meet the requirements of the service design first, but maintain a secondary focus on efficiency. When you’ve met project requirements, aggressively optimise the service to reduce costs while maintaining functionality
• your organisation must become comfortable with making regular, small scale changes to live services to make them more efficient

Digital Scotland Service Standards

Applying this principle helps you to meet the Digital Scotland Service Standards:

10. Choose the right tools and technology: the right tools and technologies are the foundation for efficient services. By making the right technology choices early on, you allow your teams to optimise and scale your services later without significant effort or disruption.

Guidance

Own your carbon footprint

Consider the environment when provisioning resources. The use of cloud services can dramatically reduce the public sector’s carbon footprint. But remember, services are still running and consuming energy somewhere.

The hyper-scale cloud providers (Microsoft, AWS and Google) have recently made commitments to achieve carbon neutrality, or carbon negativity by set dates. We will track their performance on these commitments closely, and encourage you to do the same.

Run lean and scale automatically

Cloud services offer the innate ability to scale on demand, and in many cases, transparently. Because of this, you should always provision what is required and scale automatically based on demand. Over-provisioning resources is an immediate and avoidable waste of your budget.

As most services scale automatically, capacity planning has been replaced by demand planning. The implications of traffic surges are no longer technical, but financial.

Monitor continuously

We recommend monitoring your cloud estate continuously. The major cloud vendors provide basic monitoring by default, and optionally allow you to enable more advanced and frequent monitoring.

Many of the auto-scaling methods rely on these monitoring metrics to scale your resources up and down automatically.  We recommend you enable default monitoring and auto-scaling, and configure it over time in response to changes in demand.