Cloud Principles

Whilst we have many Technical Architecture Principles, here are the key principles you should be aware of, when working with cloud.

Principle	Statement	Rationale
Cloud-First	We are a cloud-first organisation. Where solutions could equally be deployed on-premises or in public cloud, with no significant differences in cost, public cloud is preferred.	Moves us towards PAYG, rather than paying lumps of capex. Ensures deployments are automated by default. Leverages cloud scalability and elasticity, which cannot be matched on-premises. Moves us from self-managed to provider-managed services. Reduces our reliance on data centres, helping us achieve data centre exit. Leverages cloud agility.
Prefer managed platforms over DIY in cloud	Where a fully-managed service is offered in public cloud, this should be used in preference to building and self-managing a similar capability. For example, prefer using cloud-managed DB-as-a-service, over building IaaS instances and deploying a self-managed database.	Lower operational overhead for provisioning and managing; do not require manual patching; do not typically require specialist engineering capability; typically offer high availability, DR capability, and backup capability, using out-of-the-box configuration; cloud vendor-backed SLAs; typically ‘standard’ and open source technologies, under the hood.
Prefer free-and-open source (FOSS) over commercial proprietary products	Prefer an open source solution over a commercial proprietary solution, where an open source solution is established, widely adopted and has strong credentials.	Cheaper; portable; hosting-agnostic (whereas commercial solutions are often not suitable for cloud); often more secure; open standards; more attractive to internal and external talent; avoids vendor lock-in; typically better quality software with faster release cadence.
Automate	Design with automation from the start, and always build environments using Infrastructure-as-Code (IaC)	Agility; repeatable, consistent environments; avoids configuration drift; self-documenting LLD in the form of version-controlled IaC; avoids human error; can be leveraged in CI/CD pipelines.
Immutable infrastructure	Outside of sandbox environments, cloud resources should not be modified after they have been built. Instead, always make changes in the IaC and redeploy.	Same benefits as Automate
Prefer cloud-native	Where a cloud-native capability exists, it should be considered first.	Easily deployed with IaC; typically highly available, elastic, and auto-scaling out of the box; benefits from the rapid evolution of the product; reduces reliance on our own data centres; avoids complex, costly and wasteful integrations, e.g. egress of data from a cloud-hosted application back to the data centre.