Armory Continuous Deployment-as-a-Service Docs
What CD-as-a-Service is
Modern software delivery requires sophisticated control of delivery speed and traffic management, as well as integration with external tools and verification processes. Software is rarely deployed to a single environment. Most workflows require promotion from one environment or region to another after meeting constraints such as test executions, manual approvals, CI workflows, and canary analysis. CD-as-a-Service is a centralized control plane that encapsulates those features and enables deployment to multiple Kubernetes clusters using a secure Kubernetes agent architecture. CD-as-a-Service supports multi-environment, multi-region app deployments with promotion constraints and advanced deployment strategies, such as canary and blue/green.
Why you should use CD-as-a-Service
- Native Kubernetes resources: You create your Kubernetes manifests how you want - manually or by using a tool like Helm or Kustomize.
- Logicless Kubernetes agents in your clusters: CD-as-a-Service’s Remote Network Agents (RNAs) act as simple network relays and provide the CD-as-a-Service control plane with Kubernetes ServiceAccount-based credentials. You don’t need to open any ports to grant CD-as-a-Service deployment access to your Kubernetes clusters.
- Centralized business logic: Deployment logic is encapsulated in the control plane. You get new features immediately without having to worry about edge maintenance – no Remote Network Agent upgrade campaigns within your organization, no need to manage Custom Resource Definitions (CRDs).
- Declarative deployment: You define your software delivery requirements, traffic shaping, canary analysis, webhooks, and manual approvals in a declarative deployment configuration file.
- Integration with your existing SDLC: You build and publish your containers where and how you want, whether you use Docker Hub or a private registry that is only accessible in your network.
How CD-as-a-Service works
CD-as-a-Service is a centralized control plane that utilizes flexible promotion constraints to safely deploy software and/or config changes across multiple clusters, environments, and/or regions.
CD-as-a-Service uses secure, logicless Remote Network Agents to access privately networked resources and Kubernetes clusters via ServiceAccounts. You do not need to upgrade agents for new features since business logic is encapsulated in the CD-as-as-Service control plane.
Remote Network Agents connect to CD-as-a-Service to establish gRPC via HTTP/2 connections secured with TLS and OIDC client credentials. You don’t need to open any ports to grant CD-as-a-Service access to your Kubernetes clusters or privately networked resources.
- You render your Kubernets manifests using the tools you want. You build and publish your containers where and how you want, from DockerHub to a private registry on your network.
- You create a CD-as-a-Service deployment config file in which you define your deployment: canary and/or blue/green strategy; traffic shaping; deployment constraints such as manual judgments; external automation using webhooks; and retrospective analysis. You include the path to your Kubernetes manifests in the deployment config file. CD-as-a-Service can deploy any Kubernetes object that you define in your Kubernetes manifest.
- You start your deployment by sending the deployment config file to CD-as-a-Service using the Armory CLI, which you can run locally or in a Docker container. You can trigger deployments automatically from any CI system. If you’re using GitHub, Armory provides a GitHub Action for triggering deployments from your GitHub workflow.
The CD-as-a-Service control plane executes the deployments with constraints defined in your deployment config file, converting Kubernetes deployment objects into CD-as-a-Service managed ReplicaSets, handling traffic management and scaling. Remote Network Agents integrate with internal tools by securely relaying network calls and using their configurable ServiceAccount credentials to communicate with your clusters API.
Then monitor your deployment’s progress through the UI or CLI.
Was this page helpful?
Thank you for letting us know!
Sorry to hear that. Please tell us how we can improve.
Last modified July 13, 2023: (5e051be)