Skipper uses a basic client server architecture. The server exposes a REST API that is used by the interactive shell. You can brose the API using familiar HTTP client tools. The server persists Package Metadata and Release state in a relational database.
Platforms are defined using using the property prefix spring.cloud.skipper.server.platform. For each of the supported platforms, cloudfoundry, 'kubernetes’ and local, you can define multiple accounts.
Each account maps onto an instance of a Spring Cloud Deployer implementation which is responsible for
deploying the applications. The Part VI, “Installation” shows more details but it is important to note that the Skipper server is not tied to a deploying to a single platform. Wherever Skipper is running, it can be configured to deploy to any platform. For example, if Skipper is deployed on Cloud Foundry, you can still register accounts for Kubernetes and deploy apps to Kubernetes from Cloud Foundry.
The release workflow is currently a hard-coded workflow managed by the Spring Cloud State Machine project. The state of the State Machine is persisted in a relational database.