Cloud-Native Enterprise Architecture: Principles and Governance

The CNCF definition

The Cloud Native Computing Foundation (CNCF) publishes a definition of cloud native, at version 1.1. According to it, these practices help organisations develop, build and deploy workloads in public, private or hybrid clouds. They meet organisational needs at scale, in a programmatic and repeatable manner.

The systems concerned are loosely coupled and interoperate in a secure, resilient, manageable, sustainable and observable way. The definition lists, without claiming completeness, containers, service meshes, multi-tenancy, microservices, immutable infrastructure, serverless and declarative APIs. Combined with robust automation, these practices allow high-impact changes to be made frequently and predictably.

Monolith and independent services

In a monolithic application, components are released together. A local change requires redeploying the whole application, dependencies between components slow updates down and scaling applies to the entire application. A failure in one component can also stop the rest.

Cloud native splits the system into services that are released and scaled separately. Teams add a new service faster, but each split adds interfaces and dependencies to govern.

Practices to describe

A cloud-native architecture combines several practices:

• Microservices: each service is developed, deployed and scaled independently.
• Containers: the same package runs the same way from one environment to the next.
• Orchestration: the platform automates deployment, scaling, recovery and resource allocation. Kubernetes is an open source platform of this kind for containerised workloads.
• DevOps and continuous integration: releases are more frequent and checked automatically, which lowers the risk of each one.
• Infrastructure as Code: networks, machines and services are created from versioned files, reviewed like code.
• API first: each service publishes its interface contract before its implementation.

The role of enterprise architecture

Cloud native multiplies services, environments and suppliers. Enterprise architecture sets cloud standards and structures the shared platforms. It also governs APIs, tracks security and compliance and keeps microservice complexity in check.

Without this work, cloud resources multiply without an owner or a known use: this is cloud sprawl.

Expected gains and constraints

A service scales with its own load, and a failure can stay confined to one service. Teams release more often and can change technology service by service. Cost can follow actual use, provided each resource has an owner.

These gains come with constraints. Each service adds interfaces, identities, logs and costs to track. Security covers more access points, and teams need skills in cloud platforms, Kubernetes, DevOps and automation.

A gradual adoption

The transition can happen application by application. It aims at a shared platform on which teams release quickly while respecting the organisation’s standards.

• Modernise the existing applications whose target has been decided.
• Move those applications to containers.
• Set up a shared Kubernetes platform.
• Automate delivery with DevOps and continuous integration.
• Define cloud architecture governance: rules, owners of each environment and reviews.

The technology platform in ArchiLU

In ArchiLU, the technology catalogue groups components by type, including container platforms, infrastructure services and runtime environments. The technology landscape connects technologies through three relationships: hosts, depends on and succeeds.

Each application shows its technology stack. 90 days before an end of support, a case opens: extend, migrate, retire or accept the risk. Within the authorised scope, the Discovery collector observes Linux machines, Docker containers and their TCP connections; these facts become proposals for review.

Supplier contracts, cloud contracts included, are allocated to the applications they cover, with their due date and any automatic renewal flagged.