MVD deployment in a Kubernetes Cluster

[marispace-x-user]

Description

As of today, two deployments are provided within the MVD:

• an Azure deployment relying on Azure managed services (such as Azure Key vault). This related to refactor: update pipelines eclipse-dataspace-hub/MinimumViableDataspace#139.
• a local deployment relying on docker-compose with no dependency on any cloud managed service. This related to build(deps): bump edc from 0.1.2 to 0.1.3 eclipse-dataspace-hub/MinimumViableDataspace#157.

However in most cases, participant applications are running in a cluster. Kubernetes is a well-established standard in terms of container orchestration. Thus it makes sense to also propose a MVD deployment for Kubernetes.

Acceptance Criteria

• The MVD can be deployed and run on a Kubernetes cluster based on step-by-step instructions
• A data transfer can be performed between the participants, especially the possiblity for a consumer to actively query data from a data provider exposing a REST API through its Data Plane (more details can be found here). This corresponds to the typical data-exchange use-case in Catena-X/Eona-X.
• No additional vendor specific components are required. Open-source counterparts of the well-known cloud managed services are acceptable tough (e.g. Azurite)
• Keys are stored on the local file system, or in embedded secret store provided out of the box by Kubernetes
• If keys are stored in the filesystem, then documentation should contain a security warning that keys will not be securely stored, and therefore only temporary keys for development purposes shall be used
• Integration test in CI to validate the Kubernetes deployment of the MVD (note: Minikube is a good candidate here) and the data transfer between the participants

Out of scope

• Local file transfer for cloud deployments